The construct validity of the Beck Depression Inventory (BDI) in measuring treatment outcome is assessed in 103 psychiatric inpatients. In this context, construct validity means that the BDI measures the same construct in repeated measurement and that the change scores can be explained by treatment effects. In confirmatory factor analyses, only the first factor proved to be stable. In accordance with other studies, the sensitivity to therapeutic change in long-term intervals of several weeks could be confirmed. Significant changes in a short-term interval of 1 day in the non-endogenously depressed patients indicate an overreactivity of the BDI to change which cannot be explained by treatment effects or mood changes.