← Back
Phase

CFOP PLL Tutorial

Recap: from 2-look PLL to full PLL

OLL is complete — the whole top face is yellow. When you learned LBL you permuted the last layer in two looks (2-look PLL — Permutation of the Last Layer): corners first — read the headlights and use the two A-perm triggers (Aa and Ab) or the E-perm to put all four corners in their correct spots — then edges — use Ua and Ub (the 3-edge cycles) to place the remaining edges. Two separate looks.

CFOP's full PLL merges those two looks into one: read corners and edges together and solve with a single PLL algorithm. This page teaches how to recognise all 21 cases and what order to learn them — the 21 algorithms sit in the quick reference at the end of this page; drilling them one-by-one happens on the drill page.

Goal: permute the whole top layer in one look

OLL done but PLL unsolved — pieces are oriented but not permuted
All top-layer corners and edges in their correct spots = PLL solved

Done = all four top corners and all four top edges occupy their correct positions relative to the side centres — the last layer is fully permuted. A final AUF (Adjust U Face) aligns the top layer with the solved middle layer, and the cube is complete. Note the difference from OLL: here orientation (yellow-up) is already finished; only permutation — getting each piece to its correct destination — remains.

Recognition method: corners first, then edges

The key to recognising 21 PLL cases in one look is a corner-first decision tree: count headlights first (which involves only corner relationships), then inspect edge stickers for full bars and colour blocks. The tree is AUF-invariant — you do not need to normalise the U-layer before starting classification. Scan all four sides, then follow the branches below. The tree ends at a Group (5 Groups plus the PLL-skip exit), not an individual case — within-Group identification comes next.

The decision tree uses four features — the first four figures highlight where each lives; the fifth is the same T after a U turn, with all four features preserved. That is why pre-AUF does not change which case you recognize.

homepermutedsolved top
Headlights: the two corner stickers on one side share a color (middle edge ignored)
Full Bar: corner-edge-corner on one side all share a color (a full bar also counts as one set of headlights)
Color Block: an adjacent corner and edge sticker share a color (no center match required)
Connected Color Block: a color block on a side adjacent to the headlights, using one of the headlights' corners, extending from one end (T · AUF-0 baseline)
T · U: rotate the T above by U — the four blue-ring features just rotate along, unchanged; still the same case
All four sides Full Bars?
PLL-skip / AUF only
How many Headlights?
Edges second: any Color Block?
Are the Headlights a Full Bar?
Any Connected Color Block?
How many disconnected Color Blocks?

Feature → Group reference (AUF-invariant)

Recognize asGroupHeadlightsFull BarConnected Color BlockTotal Color BlocksFrequencyCases
PLL-skipsolved4481.4%
EPLLedges only40–10–215.3%Ua, Ub, H, Z
CPLLcorners only00002.8%E
CPLLcorners only100211.1%Aa, Ab
Adjacentadjacent corner swap112 or 516.7%Ja, Jb, F
Adjacentadjacent corner swap101 or 21 or 216.7%Ra, Rb, T
Diagonaldiagonal corner swap0002 or 413.9%V, Y, Na, Nb
G-permstriple corner+edge cycle100122.2%Ga, Gb, Gc, Gd
Where the frequencies come from

With F2L solved and all U-layer pieces oriented, the last layer has 288 legal permutation states (4! × 4! / 2 = 288 equal-parity corner/edge pairs). These 288 states are grouped into 22 structural orbits under AUF + viewing-rotation equivalence. One orbit is the solved/AUF-only state; the remaining 21 orbits are the named PLL cases.

Denominator 72 (unconditional). Per-case and per-group probability tables use the standard denominator 72 — they include PLL-skip (1/72 of all legal PLL states). For example, EPLL covers 11/72 ≈ 15.3% of the uniform probability space. Denominator 71 (excluding skip). The learning-stage roadmap uses coverage among solves that require a PLL algorithm, dividing by 71 instead. The two denominators serve different purposes and must not be mixed.

Learning path: build full PLL in five stages

You do not learn all 21 at once. Learn in occurrence order, scaffolding new cases against stable references. Coverage is defined as: among solves that require a PLL algorithm (excluding skips), what fraction the learner can recognise and solve directly with one algorithm. The roadmap below shows the five stages, each labelled with cumulative coverage and difficulty.

Stage 0 · EPLL + CPLL

These seven cases overlap with the existing 2-look route. Check them one by one; complete any case you cannot yet recognise and solve directly with one PLL algorithm. EPLL (Ua, Ub, H, Z) and CPLL (Aa, Ab, E) are the pure-permutation groups — only edges move, or only corners move. Learn Ua/Ub and Aa/Ab as inverse-and-mirror pairs; H, Z, and E are self-inverse.

Mastery — For each case, complete three consecutive distinct AUF variants: name the case without consulting a table, state its decisive structural cue, and execute its direct algorithm without falling back to 2-look. A wrong identification or lookup resets the count.

Stage 1 · Adjacent + Diagonal

Establish four common, approachable reference patterns before introducing their closest competitors. Adjacent (T, Jb, Ja) and Diagonal (Y). T becomes the reference for later T/R/F contrasts; Jb/Ja stay together as mirrors, both self-inverse. Y establishes the first no-headlights diagonal reference before V is introduced in Stage 2.

Mastery — For each new case, three distinct AUF variants with correct naming, cue statement, and direct execution. Then pass cumulative mixed review with Stage 0 cases — each appearing at least once, no algorithm lookups, no 2-look fallback.

Stage 2 · Adjacent + Diagonal

Add four common cases by comparing them against the stable Stage 1 references. Adjacent (Ra, Rb, F) and Diagonal (V). Compare Ra/Rb/F back to T — they share the one-headlight branch but differ by connected colour blocks. Compare V back to Y — both are zero-headlights diagonal cases, distinguished by whether the two colour blocks meet at one corner.

Mastery — Same three-AUF gate for each new case, then cumulative mixed review across Stages 0–2. No algorithm lookups, no 2-look fallback, no confusion between mirror or inverse partners.

Stage 3 · G-perms

After the simpler one-headlight branches are stable, isolate the four G-perms as one high-load block. Learn inverse pairs consecutively: Ga↔Gb, then Gc↔Gd. The group analysis shows a 2×2 relation layout: rows are inverse pairs, columns are mirror pairs. None of the four is self-inverse — be deliberate about the inverse/mirror distinction.

Mastery — Same three-AUF gate per case. Cumulative mixed review across all 17 cases from Stages 0–3. This is the hardest gate — each G-perm requires confident inverse/mirror separation.

Stage 4 · Diagonal

Na and Nb together contribute only 2/71 of non-skip PLL occurrences but carry high recognition and algorithm burden. They are self-inverse mirrors — compare both back to the earlier zero-headlights cases Y and V. Isolating them after Stage 3 preserves the useful 97% checkpoint without pretending the final two cases are easy.

Mastery — Same three-AUF gate. Final cumulative mixed review across all 21 cases. After passing, you can recognise and solve any PLL case directly — execution speed and finger tricks remain drill-page work.

Group by group: 5 Groups and 21 Cases

The 21 PLL cases cluster into five Groups by shared structural signatures. Each Group below shows its recognition-tree entry, stable landmarks, a flat gallery, within-group identification steps, inverse/mirror relationships, cross-group confusion boundaries (where applicable), and a learning-route pointer. Use these analyses to refine your recognition from Group to named case.

EPLL · 4 Cases

The four corners retain their solved relative order; only the four U-layer edges need permuting. The tree reaches EPLL when all four sides are headlights but the state is not four full bars (which would be PLL-skip). Group signature: 4 headlights; 0 or 1 full bar.

  • Four headlights distinguish EPLL from every non-skip teaching group.
  • Ua and Ub have one full bar — the three-edge cycle anchor.
  • H and Z have no full bar and no colour block.
  • On H, the middle edge on every headlights side is the colour opposite the headlights colour; on Z it is an adjacent colour.

B/R/F/L directions: the full bar (when present) supplies the orientation anchor. For H and Z, any headlights side may serve as the reference — the opposite/adjacent edge-colour comparison holds on all four sides.

homepermutedsolved top
Ua
Ub
H
Z
  1. Is there one Full Bar? → Yes → U-perm: remaining edges cycle clockwise → Ua; cycle counterclockwise → Ub.
  2. Is there one Full Bar? → No: middle edge is opposite-colour on every headlights side → H.
  3. Middle edge is adjacent-colour on every headlights side → Z.

Ua ↔ Ub are both mirrors and inverses. H and Z are self-inverse and self-mirror.

Stage 0 order: Ua → Ub → H → Z. No retained cross-group confusion after the four-headlights observation.

CPLL · 3 Cases

The four edges retain their solved relative order; only the corners need permuting. CPLL is reached through two tree entries: zero headlights + zero colour blocks (E), or one headlight + no full bar + no connected colour block + two colour blocks (Aa, Ab). Group signature: 0 headlights, 0 colour blocks (E); or 1 headlight, 0 full bars, 0 connected blocks, 2 disconnected blocks (Aa, Ab).

  • E has no headlights and no colour blocks — the only case in this category.
  • Aa and Ab have exactly one headlight and two disconnected colour blocks.
  • Using the headlights side as reference, one corner serves as the anchor around which the other three cycle.

B/R/F/L directions: mentally treat the headlights side as 12 o'clock. The two disconnected colour blocks sit on the sides adjacent to the headlights — their positions relative to the anchor corner distinguish Aa from Ab.

homepermutedsolved top
Aa
Ab
E
  1. How many Headlights? → 0, no Colour Blocks → E.
  2. 1 Headlight, two disconnected Colour Blocks → A-perm: three corners cycle clockwise around the anchor corner → Aa; counterclockwise → Ab.

Aa ↔ Ab are both mirrors and inverses. E is self-inverse and self-mirror.

CPLL ↔ Adjacent — shared one-headlight, no-full-bar first impression. Any connected colour block means Adjacent; none plus two disconnected blocks means CPLL/A. CPLL ↔ Diagonal — shared zero-headlights. Zero colour blocks means E; any colour block means Diagonal. CPLL ↔ G-perms — shared one-headlight, no full bar, no connected block. Count colour blocks: two means CPLL/A; one means G-perms.

Stage 0 order within CPLL: Aa → Ab → E.

Adjacent · 6 Cases

An adjacent pair of corners is transposed while edges are also permuted. Reached via one headlight + one full bar (Ja, Jb, F) or one headlight + no full bar + one or two connected colour blocks (Ra, Rb, T). Group signature: exactly 1 headlight; either ≥1 full bar, or ≥1 connected colour block.

  • Full-bar branch (Ja, Jb, F): Ja and Jb have three additional same-direction colour blocks outside the bar; F has none outside it.
  • No-full-bar branch (T, Ra, Rb): T has two connected colour blocks; Ra and Rb have one.
  • Mirror direction is read relative to the headlights / full-bar side, never from an absolute face.

B/R/F/L directions: centre the headlights or full-bar side at 12 o'clock. 'Lean clockwise' means the colour block uses the corner at the clockwise end of that side; 'lean counterclockwise' means the opposite end.

homepermutedsolved top
Jb
Ja
F
T
Ra
Rb
  1. Are the Headlights a Full Bar? → Yes → no Colour Block outside the bar → F.
  2. Three outside Colour Blocks, all leaning the same way → J-perm: all lean counterclockwise → Jb; all lean clockwise → Ja.
  3. Are the Headlights a Full Bar? → No → two Connected Colour Blocks → T.
  4. One Connected Colour Block → R-perm: block on the headlights' counterclockwise adjacent side → Ra; clockwise adjacent side → Rb.

Ja ↔ Jb and Ra ↔ Rb are mirror pairs whose members are each self-inverse. T and F are self-inverse and self-mirror.

CPLL ↔ Adjacent — CPLL shares one headlight and no full bar, but has no connected block and has two disconnected blocks. G-perms ↔ Adjacent — G-perms also have one headlight, but have neither a full bar nor a connected block and have only one disconnected block.

Curriculum order differs from gallery order. Stage 1: T → Jb → Ja. Stage 2: Ra → Rb → F. T is established before R/F comparisons.

Diagonal · 4 Cases

A diagonal pair of corners is transposed while edges are also permuted. Every Diagonal case has zero headlights and at least one colour block. Group signature: 0 headlights; 2 or 4 total colour blocks.

  • Y and V have two colour blocks — in V the two blocks meet at the same corner; in Y they are disjoint.
  • Na and Nb have four colour blocks, all leaning consistently around the perimeter.
  • No headlights — this immediately distinguishes Diagonal from all Groups except CPLL's E case.

B/R/F/L directions: with zero headlights, use any side as the starting reference. Scan all four sides for colour blocks and trace whether they meet at a common corner (V) or remain separate (Y). For Na/Nb, all four sides display a leaning colour block — check the consistent direction.

homepermutedsolved top
Y
V
Na
Nb
  1. How many Colour Blocks? → 2: blocks are disjoint → Y.
  2. Blocks meet at the same corner piece → V.
  3. 4 Colour Blocks: all lean counterclockwise → Na; all lean clockwise → Nb.

Na ↔ Nb are mirrors whose members are each self-inverse. Y and V are self-inverse and self-mirror.

At zero headlights, E can initially resemble a Diagonal case — but E has no color blocks at all, while every Diagonal case has two or four. One complete scan settles it.

Stage 1 introduces Y as the no-headlights reference. Stage 2 adds V. Stage 4 adds Na → Nb — compare both back to Y and V.

G-perms · 4 Cases

Three corners and three edges cycle while one corner-edge pairing stays together. Reached via one headlight + no full bar + no connected colour block + exactly one disconnected colour block. Group signature: 1 headlight; 0 full bars; 0 connected colour blocks; 1 total colour block.

  • Use the headlights side as 12 o'clock.
  • The edge between the headlights is either an adjacent colour or the opposite colour relative to the headlights colour — this is the first split.
  • The sole disconnected colour block sits either on the counterclockwise adjacent side, the clockwise adjacent side, or the side opposite the headlights — its position and lean complete the identification.

B/R/F/L directions: centre the headlights at 12 o'clock. The edge colour between the headlights (adjacent vs opposite) determines the row; the block position and lean determine the column in the 2×2 layout.

homepermutedsolved top
Ga
Gb
Gc
Gd
  1. Colour between the Headlights? → Adjacent colour → sole block on counterclockwise adjacent side → Ga; on clockwise adjacent side → Gc.
  2. Opposite colour → sole block on the side opposite the headlights: block leans counterclockwise → Gb; leans clockwise → Gd.

Inverse rows: Ga ↔ Gb and Gc ↔ Gd. Mirror columns: Ga ↔ Gc and Gb ↔ Gd. None of the four is self-inverse. The 2×2 layout is: top row Ga (adjacent, counterclockwise) and Gb (opposite, counterclockwise); bottom row Gc (adjacent, clockwise) and Gd (opposite, clockwise).

CPLL ↔ G-perms — Aa/Ab share one headlight, no full bar, and no connected block, but have two disconnected colour blocks rather than one. Adjacent ↔ G-perms — Adjacent cases have either a full bar or at least one connected colour block; G-perms have neither.

Stage 3 order: Ga → Gb → Gc → Gd, keeping inverse pairs consecutive.

Advanced: recognizing from two sides

Four-side recognition remains the baseline. The skill described here is optional: inferring a PLL case from only two adjacent visible sides (B and R), with the other two sides (F and L) hidden. This is a confidence-gated inference skill, not a replacement for the full four-side scan. When the visible evidence supports a unique case, name it; when it only narrows the field, state the candidate set; when uncertain, reveal one more side.

Practice two-side inference with Jb, Aa, and Ga
homepermutedsolved topnot observed
Jb · B/R only
Jb · all four sides

Jb (U). From B/R alone you observe: one Full Bar, plus one Colour Block continuing outside that bar, leaning counterclockwise. Full Bar + outside Colour Block → J-perm; the lean direction narrows to Jb. Revealing F/L confirms the remaining outside blocks continue in the same direction — a direct call from strong landmarks.

homepermutedsolved topnot observed
Aa · B/R only
Aa · all four sides

Aa (no AUF). From B/R alone you observe: one Headlights side, no Full Bar, one visible disconnected Colour Block. These coarse features do not reveal whether another disconnected block exists on a hidden side — they support the candidate set {Aa, Ga}. The fine cue is the edge between the headlights: relative to the headlights colour, it is the counterclockwise adjacent colour. You may propose Aa. Revealing F/L shows a second disconnected Colour Block, confirming Aa / CPLL.

homepermutedsolved topnot observed
Ga · B/R only
Ga · all four sides

Ga (U2). The same coarse description applies: one Headlights, no Full Bar, one visible disconnected Colour Block. Here the edge between the headlights is the clockwise adjacent colour — you may propose Ga. Revealing F/L confirms only one disconnected Colour Block exists globally, hence Ga / G-perms. If you cannot state the clockwise/counterclockwise colour relation confidently, the correct response is {Aa, Ga}; reveal one more side.

  1. Four-side anchoring. Review the complete Jb, Aa, and Ga views. For each, state its Group, the decisive four-side landmark, and the relationship distinguishing it from its nearest comparison.
  2. Two-side prediction and reveal. Apply the visibility mask to the same examples. Before revealing F/L, report either a unique case with supporting evidence, or a Candidate Set plus the evidence still missing. Then restore F/L and compare.
  3. Confusion-pair calibration. Alternate Aa and Ga. State their shared coarse evidence, read the clockwise/counterclockwise adjacent colour between the headlights, make a supported proposal, request a third side when the fine relation is not stable, then reveal and verify the total disconnected-block count.
  4. Transfer to mixed practice. For any cases already completed in the main route: inspect the current adjacent two sides first, state a unique Case or Candidate Set before rotating, reveal only one additional side if necessary, and finish with a four-side structural review.

Theory: why there are 21 PLL Cases

Behind the 21 case pictures are only a few kinds of piece cycles. The three folds below build a ladder: first a readable cycle vocabulary for the five Groups, then a proof that the 288 legal PLL states collapse to 21 named cases, and finally a look at how a handful of generating operations span the entire PLL state space — and what that tells us about inverse, mirror, and algorithm relationships.

From five groups to cycle types

Each PLL Group has a characteristic pair of cycle types — one describing how the four corners move, the other describing the four edges. Introduce just four structures: 3-cycle (three pieces cycle, the fourth stays — two swaps, even parity); double transposition (two disjoint pairs swap — two swaps, even); single transposition (one pair swaps — odd); and fixed (no movement for that piece type).

GroupCorner cycleEdge cycleCasesMeaning
EPLLnone3-cycle or double-swapUa, Ub; H, Zedges only
CPLL3-cycle or double-swapnoneAa, Ab; Ecorners only
Adjacentone adjacent corner swapone swapT, Ja, Jb, F, Ra, Rbadjacent corner pair + edge pair swap together
Diagonalone diagonal corner swapone swapV, Y, Na, Nbdiagonal pair + edge pair swap together
G-perms3-cycle3-cycleGa, Gb, Gc, Gdthree corners + three edges cycle around a fixed corner-edge block

Every row has matching corner and edge parity — this is the readable reason a legal cube cannot swap just two corners or just two edges.

Why there are 21 PLL cases

Assume F2L solved, all U-layer pieces oriented, only the four corners and four edges permuted. Index the four U-layer slots and solved labels clockwise 0, 1, 2, 3. A fixed-frame state is a pair (c, e) where c, e ∈ S₄ are the corner and edge arrangements, with equal parity (sign(c) = sign(e)). Each S₄ has 12 even and 12 odd permutations, so

|X| = 12×12 + 12×12 = 288

Two states share the same named PLL structure when they differ only by pre-AUF, target-frame/final-AUF equivalence, or whole-cube y rotation — together an action of H = C₄ × C₄ on X.

Burnside's lemma counts the orbits: identity (0,0) fixes 288; the four 4-cycle pairs (a,b ∈ {1,3}) each fix 8 states; the single (2,2) pair fixes 32; the 10 mismatched-cycle-type pairs fix 0.

(288 + 4×8 + 32) / 16 = 352 / 16 = 22 orbits

One is the solved state → 21 non-solved PLL cases.

Do not quotient inverse or mirror — Aa and Ab, Ua and Ub, Na and Nb, and all four G-perms remain distinct named cases. The frequency table (denominator 72) follows from the orbit-size distribution: 16 orbits of size 16, 2 of size 8, and 4 of size 4 (including skip).

Generators and algorithm relationships

Choose fixed-frame representatives: α = Aa (corner 3-cycle, edges fixed), ε = Ua (edge 3-cycle, corners fixed), r = U (simultaneous 4-cycle on corners and edges). Their closure ⟨α, ε, r⟩ = X, with ⟨Aa, Ua, U⟩ = 288.

Proof sketch: conjugates rᵏαr⁻ᵏ move the corner 3-cycle to different triples, generating all even corner permutations (A₄). Conjugates rᵏεr⁻ᵏ similarly generate all even edge permutations (A₄). Together they reach the even-even subgroup A₄ × A₄ (144 states). The 4-cycle r is odd, reaching the odd-odd coset (144 more states) → 288 total. This is a mathematical generating set, not a recommended solve method — never execute Aa and Ua repeatedly in a timed solve instead of learning one-step PLL algorithms. Theory relates the cases structurally but does not choose finger tricks — each Case's actual algorithm is in the quick reference at the end of this page.

ConjugateX A X⁻¹ moves the work area, performs the core effect, moves back. e.g. UᵏAU⁻ᵏ relocates the corner 3-cycle to a different set of three corners. Commutator[A,B] = A B A⁻¹ B⁻¹. When A and B overlap locally, the cancellation can leave a small net cycle — this supplies intuition for longer algorithms that temporarily disturb many pieces.

Inverse: Aa→Ab, Ua→Ub, Ga→Gb, Gc→Gd; E, F, H, Ja, Jb, Na, Nb, Ra, Rb, T, V, Y, Z are self-inverse. Mirror: Aa→Ab, Ua→Ub, Ja→Jb, Ra→Rb, Na→Nb; inverse(Ga) = Gb while mirror(Ga) = Gc — a decisive contrast that prevents the common mistake of assuming mirror equals inverse. Mirror relations describe case geometry, not finger-trick implementation.

These elementary structures (3-cycles, double transpositions, coordinated corner/edge transpositions) explain why algorithms cluster and why setups, inverses, and mirrors reuse ideas. They do not uniquely determine a move sequence — many algorithms realise the same permutation with different grips, rotations, and intermediate disturbance.

Theory tells you what kind of permutation a Case performs and why related Cases reuse structures. It does not choose the best finger trick for you. Each Case's exact algorithm is in the quick reference at the end of this page; the breakdown and execution details are drilled on practice.

Common recognition traps

The usual PLL recognition pitfalls to watch for:

  • Confusing Group with Stage. A Group is a structural family (EPLL, CPLL, Adjacent, Diagonal, G-perms). A Stage is a learning-sequence bucket (0–4). They do not correspond one-to-one; e.g. Adjacent cases appear across Stages 1 and 2, and Diagonal appears in Stages 1, 2, and 4.
  • Incomplete four-side scan. Stopping after checking only two or three sides can miss a headlights side, a full bar, or a connected colour block. Always scan all four sides before entering the tree.
  • Headlights vs Full Bar. A Full Bar is a stronger landmark than Headlights alone, but it still counts as one set of headlights — do not treat it as a separate measurement standard. The tree answers 'how many headlights?' first, then checks whether that headlight side is also a full bar.
  • Centre-match vs same-colour. A corner sticker matching another corner sticker (a headlights pair) is not the same as matching the side centre. The tree relies only on corner-to-corner and corner-to-edge colour relationships — centre matching is relevant only for the final AUF, not for case classification.
  • Inverse vs mirror, especially G-perms. inverse(Ga) = Gb, mirror(Ga) = Gc — these are different relationships. Do not assume that mirrored cases are also inverses, or that a case you know in one orientation can be mirrored simply by performing the same algorithm on the opposite side.
  • Unsupported two-side guesses. Two-side inference is optional and confidence-gated. If you cannot name a unique case or a small candidate set from two sides, reveal a third side — do not guess. Guessing from insufficient evidence is not PLL recognition.
  • Missing the final AUF. The PLL algorithm permutes the pieces relative to each other, but may leave the top layer misaligned with the side centres. A final AUF (U / U2 / U') is part of every PLL solve — forgetting it means the cube is not solved.

Go drill the 21 PLL Cases

Algorithm quick reference · all 21
homepermutedsolved top
EPLL · 4 Cases · 11/72
Ua
M2 U M U2 M' U M2
Ub
M2 U' M U2 M' U' M2
H
M2 U M2 U2 M2 U M2
Z
M' U M2 U M2 U M' U2 M2 U'
CPLL · 3 Cases · 10/72
Aa
x' R2 D2 R' U' R D2 R' U R'
Ab
x' R U' R D2 R' U R D2 R2 x
E
x' R U' R' D R U R' D' R U R' D R U' R' D' x
Adjacent · 6 Cases · 24/72
Jb
R U R' F' R U R' U' R' F R2 U' R' U'
Ja
z U' R D' R2 U R' U' R2 U D R'
F
R' U' F' R U R' U' R' F R2 U' R' U' R U R' U R
T
R U R' U' R' F R2 U' R' U' R U R' F'
Ra
R U' R' U' R U R D R' U' R D' R' U2 R' U'
Rb
R' U2 R U2 R' F R U R' U' R' F' R2 U'
Diagonal · 4 Cases · 10/72
Y
F R U' R' U' R U R' F' R U R' U' R' F R F'
V
R' U R U' R' f' U' R U2 R' U' R U' R' f R
Na
R U R' U R U R' F' R U R' U' R' F R2 U' R' U2 R U' R'
Nb
R' U R U' R' F' U' F R U R' F R' F' R U' R
G-perms · 4 Cases · 16/72
Ga
R2 U R' U R' U' R U' R2 U' D R' U R D' U
Gb
F' U' F R2 u R' U R U' R u' R2
Gc
R2 u' R U' R U R' u R2 f R' f'
Gd
R U R' F2 D' L U' L' U L' D F2

This page gives the recognition method + learning path + Group analysis for full PLL. The 21 algorithms are in the quick reference above, grouped by Group for lookup as needed; drilling them to speed happens on the drill page. Accuracy first: do not move to the next stage until the cumulative mixed review passes within the current coverage.

Drill all 21 PLL Cases