Sample-based profiling is the practice of using a statistically representative subset of all possible data, or samples, to profile an application. If done right, this usually leads to much less profiling overhead and better information than alternative techniques.

See also Exact profiling.