Racial (black-white) comparisons were made on four dependent measures (MMPI) to probe three issues generated by previous inconsistent findings in this area. Issues were (a) defining primary sources of variance, (6) comparability of dependent measures, and (c) effect of selection criteria, in the form of common requirements for profile validity, on outcome. White counterparts were matched with 160 blacks on sex, socioeconomic status (education and occupation), hospital status, age, and duration of illness. Phase 1 analysis (# = 320) used all profiles. Phase 2 analysis (# = 116) was limited to valid profiles. Results showed (a) race to be a primary source of variance, (b) dependent measures are not entirely comparable, and (c) application of selection criteria is a powerful parameter influencing outcome. Blacks elevated Scales F, 1, 5, 8, 9, and the overall profile mean, more often elevated Scale 8 as the highest single point code, Scale 1 as the second-highest code, and produced more 8-6 and 2-4 codes than whites. Whites elevated Scale 3, more often elevated Scale 7, as the highest and as the second-highest point code, and produced 2-7 and 4-7 codes more frequently than did blacks.