It remains uncertain whether bipolar disorder (BPD) patients in randomized-controlled trials (RCTs) are sufficiently representative of clinically encountered patients as to guide clinical-therapeutic practice. We complied inclusion/exclusion criteria by frequency from reports of 21 RCTs for mania, and applied them in a pilot study of patients hospitalized for DSM-IV BPD manic/mixed states to compare characteristics and clinical responses of patients who did versus did not meet exclusion criteria. From 27 initially identified inclusion/exclusion criteria ranked by citation frequency, we derived six inclusion, and 10 non-redundant-exclusion factors. Of 67 consecutive patients meeting inclusion criteria, 15 (22.4%) potential “research subjects” met all 10 exclusion criteria. The remaining 52 “clinical patients” differed markedly on exclusion criteria, including more psychiatric co-morbidity, substance abuse, involuntary hospitalization, and suicide attempts or violence, but were otherwise similar. In both groups responses to clinically determined inpatient treatments were similar, including improvement in mania ratings. Based on applying reported inclusion/exclusion criteria for RCTs to a pilot sample of hospitalized-manic patients, those likely to be included in modern RCTs were similar to patients who would be excluded, most notably in short-term antimanic-treatment responses. The findings encourage further comparisons of subjects included/excluded from RCTs to test potential clinical generalizability of research findings. The pilot study is limited in numbers and exposure times with which to test for the minor differences between “research subjects” and “clinical patients.” Copyright © 2008 John Wiley & Sons, Ltd.