The agency recommends that sponsors can use analysis of covariance (ANCOVA) to adjust for differences between treatment groups in relevant baseline variables, but should not use ANCOVA to adjust for variables that might be affected by treatment. (Regulatory Focus)