HIV infection attributable to medical injections is suspected to be low, although case-control studies have not provided definite results. This study aims to determine the number of HIV infections caused by the reuse of syringes and needles in the Mbeya Region, Tanzania. The direct identification or detection of HIV in syringes and needles under field conditions was not appropriate, therefore a surrogate marker consisting of two components for possible HIV transmission was used: insufficient sterilization, and blood remaining from a previous patient. The assumption was that HIV infection can only occur if both markers are positive. Samples were collected in nine health-care facilities. All syringes and needles prepared for use in these facilities were collected without prior notification. The samples were rinsed and the resulting fluid was cultured for bacteria. Traces of blood were detected by urine stick test for haemoglobin volumes > 0.0015 microliters. Bacterial contamination was found in 32.8% of the total 1219 syringes and needles; 67% was caused by improper handling of the equipment after sterilization. Blood was detected in 12.5% of the samples. In the following three sampling strata, both contamination criteria were positive either on the syringe or the needle: wards/outpatient departments (OPD), 1.39%; laboratories, 7.45%; expanded programme on immunization (EPI), < 0.1%. We calculated that from 1.1 million patients injected in wards/OPD in any 1 year, fewer than 13 become infected, in laboratories fewer than 12 (160,000 blood-taking procedures), and less than one child in the EPI (850,000 vaccinations). With an established AIDS intervention programme supporting the health system, less than 0.4% of the total annual incidence of 4500-8500 is attributable to medical injections in the Mbeya Region.