Cross cultural research on substance use disorders (SUD) demands diagnostic measures and criteria that apply equally well to persons of different ethnic backgrounds. To evaluate the reliability of SUD in different ethnic groups, comparisons were made of the one week test/retest agreement on DSM-IV lifetime dependence disorders for 196 African-American (AA) and 107 Caucasian (C) respondents using the Composite International Diagnostic Interview-Substance Abuse Module (CIDI-SAM). Overall we found excellent reliability, using kappa (k) statistics, in diagnosing both AA and C respondents with alcohol dependence (AA k = 0.78; C k = 0.80) and opiate dependence (AA k = 0.77; C k = 0.71), good reliability for diagnosing both AA respondents (k = 0.63) and C respondents with cocaine dependence (k = 0.67), and to good reliability for both AA and C respondents with cannabis dependence (AA k = 0.50; C k = 0.69). Reliability of the dependence/abuse criteria was consistent with the overall diagnostic reliability but some variation was noted. No significant differences in the kappas were found between the two ethnic groups for any of the substance dependence diagnoses, and only one dependence or abuse criterion (continued use of cocaine despite physical/psychological problems) differed significantly between AA and C respondents. These initial results indicate that DSM-IV dependence diagnoses as measured by the CIDI-SAM apply equally well to AA and C respondents.