Objective To assess prevalence, disease stage, and linkage to HIV care following diagnosis at a mobile HIV testing unit, compared to clinic-based testing, in a Durban township. Design Prospective cohort study. Methods We enrolled adults presenting for HIV testing at a community-based mobile testing unit (mobile testers) and at an HIV clinic (clinic testers) serving the same area. Testers diagnosed with HIV, regardless of testing site, were offered immediate CD4 testing and instructed to retrieve results at the clinic. We assessed rates of linkage to care, defined as CD4 result retrieval within 90 days of HIV diagnosis and/or completion of ART literacy training, for mobile versus clinic testers. Results : From July-November 2011, 6,957 subjects were HIV tested (4,703 mobile and 2,254 clinic); 55% were female. Mobile testers had a lower HIV prevalence than clinic testers (10% versus 36%), were younger (23 versus 27 years) and were more likely to live >5 km or >30 minutes from the clinic (64% versus 40%; all p< 0.001). Mobile testers were less likely to undergo CD4 testing (33% versus 83%) but more likely to have higher CD4 counts (median 416/μl, IQR 287-587 versus 285/μl, IQR 136-482) than clinic testers (both p<0.001). Of those who tested HIV positive, 10% of mobile testers linked to care, versus 72% of clinic testers (p <0.001). Conclusions Mobile HIV testing reaches people who are younger, more geographically remote, and with earlier disease compared to clinic-based testing. Fewer mobile testers underwent CD4 testing and linked to HIV care. Enhancing linkage efforts may improve the impact of mobile testing for those with early HIV.