The aim of this study was to investigate whether multisite sampling increased heterogeneity among a sample of cocaine users from São Paulo, Brazil. Six hundred and fourteen cocaine users were interviewed at 23 fixed sites plus an out-of-treatment sample. The sites were then regrouped into six main types: university outpatient clinics, public outpatient clinics, public inpatient units, private inpatient units, HIV services and non-treatment. Marked differences were found between users recruited at these sites, especially in relation to age, gender, employment status, criminal history, history of prostitution, previous drug misuse treatment, duration of cocaine use and lifetime use of intravenous cocaine. These results suggest that multisite sampling is a valid method for increasing patient heterogeneity, but whether it improves representativeness and thus the generalisability of drug misuse research is debatable.