Principal component (PC) analysis is a statistical technique that has been used to identify core depression symptoms on the Hamilton Rating Scale for Depression (HAM-D). PC analysis is also a useful method to identify unidimensional scales of the HAM-D that are more sensitive to change following antidepressant treatment. Although there have been previous PC investigations of various versions of the HAM-D, there have been no investigations of the 31-item HAM-D scale or investigations that include subjects administered bupropion SR. We performed a PC analysis on data from 910 outpatients who participated in randomized, double-blind trials evaluating bupropion SR versus placebo for major depression. The goal of our analysis was to 1) identify components (domains) of the 31-item HAM-D and 2) determine patient response to bupropion SR using the domains identified. PC analysis produced a solution comprised of 7 domains of the HAM-D that accounted for approximately 49% of the total variance. Bupropion SR demonstrated a significant reduction (p<.01, least square mean change) in symptoms over placebo on 4 domains (cognitive, retardation, fatigue/interest, and anxiety items).