Code quality · Updated 2026-09-17
Cyclomatic complexity
A count of the independent paths through a piece of code — effectively the number of branch points plus one.
What it is
Every if, for, case and boolean operator adds a path. The metric counts them, which makes it a decent proxy for how many test cases full branch coverage would need — its original purpose.
What it is good for
Finding the outliers. A function scoring 40 in a codebase where the median is 4 is worth looking at, not because the number is bad but because something unusual is happening there. As a sorting mechanism for where to spend refactoring effort, it works.
What it is bad for
Being a gate. A switch statement mapping twenty enum values scores terribly and is perfectly readable; a deeply nested three-branch function with confusing names scores well and is not. Complexity thresholds enforced in CI mostly teach people to split functions arbitrarily to get under the limit.
Cognitive complexity was designed to address exactly this gap, by weighting nesting more heavily and not punishing flat structures.
Why it matters when you are evaluating
Most code quality platforms report it. Treat it as navigation, not judgement, and be suspicious of any tool that converts it directly into a grade.
Common mistakes
- Setting a hard threshold and enforcing it across a legacy codebase.
- Reading a low score as evidence of readable code.
- Comparing raw scores across languages, where the same construct costs different amounts.