The Timeline Followback (TLFB), a retrospective calendar-based measure of daily substance use, was initially developed to obtain self-reports of alcohol use. Since its inception it has undergone extensive evaluation across diverse populations and is considered the most psychometrically sound self-report measure of drinking. Although the TLFB has been extended to other behaviors, its psychometric evaluation with other addictive behaviors has not been as extensive as for alcohol use. The present study evaluated the test-retest reliability of the TLFB for cocaine, cannabis, and cigarette use for participants recruited from outpatient alcohol and drug treatment programs and the general community across intervals ranging from 30 to 360 days prior to the interview. The dependent measure for cigarette smokers and cannabis users was daily use of cigarettes and joints, respectively, and for cocaine users it was a "Yes" or "No" regarding cocaine use for each day. The TLFB was administered in different formats for different drug types. Different interviewers conducted the two interviews. The TLFB collected highly reliable information about participants' daily use of cocaine, cannabis, and cigarettes from 30, 90, to 360 days prior to the interview. Findings from this study not only suggest that shorter time intervals (e.g., 90 days) can be used with little loss of accuracy, but also add to the growing literature that the TLFB can be used with confidence to collect psychometrically sound information about substance use (i.e., cocaine, cannabis, cigarettes) other than alcohol in treatment- and nontreatment-seeking populations for intervals from ranging up to 12 months prior to the interview.