OBJECTIVES: To estimate the impact of injecting drug use (IDU) on mortality in HIV-infected patients in the highly active antiretroviral therapy (HAART) era. DESIGN: Population-based, nation-wide prospective cohort study in Denmark (the Danish HIV Cohort Study). METHODS: A total of 4578 HIV-infected patients were followed from 1 January 1997 or date of HIV diagnosis. We calculated mortality rates stratified on IDU. One-, 5- and 10-year survival probabilities were estimated by Kaplan-Meier methods, and Cox regression analyses were used to estimate mortality rate ratios (MRR). RESULTS: Of the patients, 484 (10.6%) were categorized as IDUs and 4094 (89.4%) as non-IDUs. IDUs were more likely to be women, Caucasian, hepatitis C virus (HCV) co-infected and younger at baseline; 753 patients died during observation (206 IDUs and 547 non-IDUs). The estimated 10-year survival probabilities were 53.2% [95% confidence interval (CI): 48.1-58.3] in the IDU group and 82.1% (95% CI: 80.7-83.6) in the non-IDU group. IDU as route of HIV infection more than tripled the mortality in HIV-infected patients (MRR: 3.2; 95% CI: 2.7-3.8). Adjusting for potential confounders did not change this estimate substantially. The risk of HIV-related death was not increased in IDUs compared to non-IDUs (MRR 1.1; 95% CI 0.7-1.7). CONCLUSIONS: Although Denmark's health care system is tax paid and antiretroviral therapy is provided free of charge, HIV-infected IDUs still suffer from substantially increased mortality in the HAART era. The increased risk of death seems to be non-HIV-related and is due probably to the well-known risk factors associated with intravenous drug abuse.