Statistical methods for evaluating the effects of treatments and prognostic factors in clinical trials are discussed. Exploratory data analysis, nonparametric methods, regression modeling, and regression diagnostics of influential cases are applied to the analysis of a pilot 'randomized' controlled trial on the treatment of acute stroke with acupuncture. The utility of this analysis for modifying patient eligibility criteria, determining required sample size and utilizing stratified randomization in a future extensive stroke trial is discussed.