CFOP: overview and learning order

CFOP is the method used by the vast majority of speedcubers. The name is an acronym of its four steps: Cross, F2L, OLL, PLL. It’s also called the Fridrich method, after Jessica Fridrich, who popularised it in the 1980s. The idea: solve the first two layers at the same time, then finish the last layer in exactly two algorithms.

The four steps

1. Cross

The four white edges around the white centre, each one lined up with the centre of its second colour. No algorithms here: it’s all thinking. A good cuber does it in six to eight moves, planned before starting, with the cube held white on the bottom. More in The cross.

2. F2L, the first two layers

For each of the four slots, you insert the white corner and the edge that goes above it together. The two pieces form a pair that goes in all at once. There are 41 starting situations, but you first learn to solve them by understanding them, without algorithms. It’s the longest step of a solve, and the one where you improve the most. More in F2L.

3. OLL, orienting the last layer

The whole top face turns yellow in one algorithm, without touching the first two layers. 57 cases, recognised by the shape the yellow stickers make. You start with a 2-look version (10 algorithms) before learning all 57. More in OLL.

4. PLL, permuting the last layer

The last-layer pieces move to their places in one algorithm: the cube is solved. 21 cases, recognised by the headlights and blocks on the sides. Here too, a 2-look version (6 algorithms) lets you start straight away. More in PLL.

1 · Cross

2 · F2L

3 · OLL

4 · PLL

By the numbers

Step	Cases	Algorithms to start with	Typical moves
Cross	–	0 (intuitive)	6 to 8
F2L	41	0 (intuitive), then refine	25 to 35
OLL	57	10 (2-look), then 57	8 to 12
PLL	21	6 (2-look), then 21	10 to 15

A full CFOP solve takes roughly 55 to 60 moves, against a hundred or more for a beginner method. More importantly, it needs fewer looks: four or five moments where you work out what to do, instead of about twenty.

What actually saves you time

In order of importance, not in step order:

1. F2L without pauses. It’s half the solve. Spotting the next pair while you insert the current one (look-ahead) is worth more than every algorithm in the world.
2. Recognition. Knowing which case is in front of you in a split second, for OLL as well as PLL. That’s what the whole Recognising cases section trains.
3. Full PLL, all 21. One algorithm instead of two at the end of every solve, and fast ones at that.
4. A planned cross during inspection, flowing straight into the first pair.
5. Full OLL, all 57. Useful, but they come last: 2-look OLL only costs you two or three seconds.

Recommended learning order

1. Cross on the bottom, without looking underneath.
2. Intuitive F2L, until you no longer hesitate on any pair.
3. 2-look OLL (10 algorithms).
4. 2-look PLL (6 algorithms).
5. The remaining 15 PLLs, by frequency.
6. Look-ahead in F2L and fingertricks; start timing your solves.
7. The remaining 47 OLLs, by shape family.
8. The 41 named F2L cases, to fix the ones you solve badly.

If you’re coming from the beginner method, From the beginner method to CFOP goes through each of these stages in detail. The learning plan spreads them over weeks.

In the app

Pick “Advanced · CFOP” in the method selector in the header, or “Intermediate · 2-look” to start with the last layer in four looks and sixteen algorithms (the same ones as in CFOP, which you’ll only need to add to later). Journey takes you through the steps, and within each step the cases are shown on the virtual cube with their algorithm animated. The Formulas sheet gathers all four steps and every case on one printable page.