The last layer is where most people get stuck. The first two layers feel logical, then suddenly you're staring at a scrambled top and reaching for a tutorial. The fix is a clear two-part system: first make the top one color, then slide the pieces into place.
This picks up after you can already solve the first two layers. If you're not there yet, start with the beginner's method, then come back.
The short version
- The last layer has two jobs: orient (OLL) so the top is one color, then permute (PLL) so pieces line up.
- The "2-look" approach splits each job in two, so you learn about 16 algorithms instead of 78.
- Recognition — spotting which case you have — is the slow part, not the finger work.
- Drill each case until you see it in under a second.
What are OLL and PLL?
OLL and PLL are the two halves of the last layer. OLL (Orientation of the Last Layer) turns every top piece the right way up, so the whole top face is one solid color. PLL (Permutation of the Last Layer) then moves those already-oriented pieces into their correct spots.
Do them in that order — orient first, permute second — and the cube finishes. Full OLL is 57 algorithms and full PLL is 21, but you don't need all 78 to solve the last layer reliably.
Step 1: 2-Look OLL (make the top one color)
2-look OLL splits orientation into two easy passes. First you orient the edges to make a cross on top (a few small cases), then you orient the corners so the whole face is one color (seven cases). That's about ten algorithms for every possible orientation.
Start with just the edge cases and one or two corner cases, and use them repeatedly — you can always fall back to repeating a short algorithm until the case becomes one you know. Full OLL comes much later, once you're consistent.
Step 2: 2-Look PLL (line everything up)
2-look PLL also splits into two passes: permute the corners first, then the edges. It uses about six algorithms while you work toward the full 21. Once corners are placed, a single edge-cycle algorithm usually finishes the solve.
If sequences like R U R' U' R' F R2 U' R' U' R U R' F' still look like noise, keep the notation cheat sheet open. Every algorithm is just those moves in order.
Why recognition matters more than speed
The slow part of the last layer isn't your hands — it's recognizing which case you're looking at. A fast solver spots the OLL or PLL pattern in a fraction of a second, then runs the algorithm on muscle memory without thinking.
So drill for recognition, not just execution. Look at the case, name it, then perform it — over and over — until the two happen almost together. Cuby's PLL trainer shows each of the 21 cases and times how long you take to recognize them, so you can target the ones that trip you up.
How many algorithms do you actually need?
For a dependable last layer, about sixteen: roughly ten for 2-look OLL and six for 2-look PLL. That's a manageable set, and it already gets many solvers well under 30 seconds.
Expand from there only when you want to. Learn all 21 PLLs next (the best return), then full OLL over a few months. For the bigger speed picture — turning, look-ahead, and F2L — see how to get faster at solving a Rubik's Cube and the CFOP method explained.
Frequently asked questions
What's the difference between OLL and PLL?
OLL orients the last-layer pieces so the top face is one color; PLL then permutes those pieces into their correct positions. You always orient first and permute second. Full OLL is 57 algorithms and full PLL is 21, but the 2-look versions need far fewer.
How do I learn the last layer without memorizing 78 algorithms?
Use the 2-look method. Split orientation into edges then corners, and permutation into corners then edges. That's roughly 16 algorithms total instead of 78, and it handles every last-layer case — just with an extra step compared to the one-look versions.
Why do I keep getting stuck on the last layer?
Usually it's a recognition problem: you don't yet know which case you're seeing, so you can't pick the right algorithm. Drilling each case with a trainer — looking, naming, then solving — builds the pattern recognition that makes the last layer feel automatic.
Should I learn OLL or PLL first?
Learn 2-look PLL alongside a simple 2-look OLL, but if you only have time for one, prioritize PLL — those 21 cases give the biggest speed return of any algorithm set in the method.
Drill it until it's automatic
The last layer stops being scary the moment you split it into orient, then permute. Learn 2-look OLL and PLL, drill for recognition, and let the patterns become instant.
Open the PLL trainer in Cuby → to study every case and time your recognition — free, in your browser.