This study compares ratings obtained with an itemized, clinician-rated, symptom severity measure--the 24-item Brief Psychiatric Rating Scale (BPRS(24))--to a Physician Global Rating Scale (PhGRS), a Patient Global Rating Scale (PtGRS) and the clinician-completed Multnomah Community Ability Scale (MCAS) in patients with bipolar disorder (BPD). A total of 69 patients (25 inpatients and 44 outpatients) with BPD were enrolled in a feasibility study of the use of medication algorithms in the treatment of BPD. Clinicians at each visit completed the BPRS(24), PhGRS and MCAS, and patients completed the PtGRS. Analyses compared the BPRS(24) and BPRS subscales with the PtGRS, PhGRS and MCAS. PtGRS scores correlated poorly with BPRS(24) and with PhGRS scores at baseline, although PtGRS change scores correlated moderately with BPRS(24) change scores. Baseline BPRS(24) and PhGRS scores correlated moderately at baseline with somewhat stronger correlations found on change scores for the two measures. MCAS scores showed moderate correlations with BPRS(24) scores both at baseline and with change over time. Global assessments by patients or physicians only moderately or poorly reflected BPRS(24) scores. Itemized symptom measures to gauge severity of illness or change over time are preferred over patient or physician global judgments.