Abstract
Methodological difficulties in outcome research in drug therapy are reviewed. No single factor has been identified as a reliable predictor of drug response, and it is unlikely that such a single predictor will prove useful in a heterogenous illness such as schizophrenia. (Abstract amended)