The CATIE (Clinical Antipsychotic Trials of Intervention Effectiveness) schizophrenia study sought to compare the effectiveness and cost-effectiveness of four second-generation antipsychotics and one first-generation antipsychotic in the treatment of schizophrenia. This study design posed several challenges for statistical analysis. The authors describe the stratified Phase 1/1A randomization, and explain the steps for comparing treatment groups within the stratified randomization structure. They describe strategies to perform treatment group comparisons that control the inflation of Type 1 error due to multiple pair-wise testing, and focus on the evaluation of multiple outcomes. The authors examine the advantages of using all-cause treatment discontinuation as the primary effectiveness outcome and the specific statistical issues for its analysis, and address the impact of missing data due to phase discontinuation on analysis of the secondary outcomes. The authors contrast the statistical methods employed to address this issue, and consider further methods.