Patient selection and dropout rates can affect the results of a clinical trial. Long lists of exclusions in the selection of patients for clinical trials reduce the possibility of examining treatment responses for heterogeneity and make recruitment difficult. In many cases, a pool of 100 potential subjects may yield only 2 or 3 qualified participants, a fact that raises the issue of generalizability of results. Dropouts should be carefully defined in advance and can be used as dependent variables for the comparison of different treatments. This article discusses some of the sampling characteristics (gender, age, diagnosis, inpatient/ outpatient status, prior neuroleptic use, and symptom severity) and dropout rates in 5 recent comparative clinical trials of atypical antipsychotics.