This paper uses data from the Global Drug Survey to test the hypothesis that there is social bias in the policing of drug users, in the form of stop and search for drugs. The Global Drug Survey is a self-report, internet survey. In 2012, it included a non-random sample of illicit drug users in the UK (n=5,919) and Australia (n=5,707). We discuss previous research on social bias in policing. We argue that an intersectional approach is necessary in order to analyse patterns of stop and search for drugs across drug users who have various social statuses. In order to analyse the influence of various patterns of drug use, we create an inductive typology of a wide range of drug use types and temporalities, using latent class analysis. We use these latent classes, frequency of past month drug use and indicators of drug dependency alongside sociodemographic variables in binary logistic regression analyses of the odds of reporting being stopped and searched for drugs in the past year. We use these models to test both consensus and conflict perspectives on the policing of drug users. We find support for both perspectives in both countries. Patterns of drug use do significantly predict the odds of sampled drug users reporting police stop and search, as expected by the consensus perspective. But drug users who were younger, male and of less advantaged social status (as measured by education in the UK sample, and by minority ethnicity, income and unemployment in the Australian sample) also had significantly higher odds of reporting stop and search. This supports the conflict perspective on policing and our hypothesis that there is evidence of social bias in the policing of drug users in the Global Drug Survey sample.