The probability that an experiment will detect a true treatment effect if one exists; depends on sample size, effect size, and outcome variance.