This paper collects a unique data set of classified ads and exploits quasi-random variation in the applicant pool composition to test for hiring discrimination against immigrants in South Africa's informal sector. Consistent with a tournament models in which immigrants are penalized, the analysis finds that foreigners and natives benefit from being pooled with foreign job seekers. Next, the paper tests whether discrimination affects search behavior. Controlling for location fixed effects, the analysis finds suggestive evidence for sorting: immigrants search further away and higher discrimination in the residential area is positively correlated with the decision to search in different suburbs.