BACKGROUND: There have been no studies showing that participation in programmes which provide legal access to drug-injection equipment leads to individual-level protection against incident HIV infection. We have compared HIV incidence among injecting drug users participating in syringe-exchange programmes in New York City with that among non-participants. METHODS: We used meta-analytic techniques to combine HIV incidence data from injecting drug users in three studies: the Syringe Exchange Evaluation (n = 280), in which multiple interviews and saliva samples were collected from participants at exchange sites; the Vaccine Preparedness initiative cohort (n = 133 continuing exchanges and 188 non-exchangers, in which participants were interviewed and tested for HIV every 3 months; and very-high-seroprevalence cities in the National AIDS Demonstration Research (NADR) programme (n = 1029), in which street-recruited individuals were interviewed and tested for HIV every 6 months. In practice, participants in the NADR study had not used syringe exchanges. FINDINGS: HIV incidence among continuing exchange-users in the Syringe Exchange Evaluation was 1.58 per 100 person-years at risk (95% CI 0.54, 4.65) and among continuing exchange-users in the Vaccine Preparedness Initiative it was 1.38 per 100 person-years at risk (0.23, 4.57). Incidence among non-users of the exchange in the Vaccine Preparedness Initiative was 5.26 per 100 person-years at risk (2.41, 11.49), and in the NADR cities, 6.23 per 100 person-years at risk (4.4, 8.6). In a pooled-data, multivariate proportional-hazards analysis, not using the exchanges was associated with a hazard ratio of 3.35 (95% CI 1.29, 8.65) for incident HIV infection compared with using the exchanges. INTERPRETATION: We observed an individual-level protective effect against HIV infection associated with participation in a syringe-exchange programme. Sterile injection equipment should be legally provided to reduce the risk of HIV infection in persons who inject illicit drugs.