In August, 1993, 13 dialysis patients at one dialysis centre in Colombia, South America, were found to be HIV positive, and this prompted an epidemiological investigation. We carried out a cohort study of all dialysis centre patients during January, 1992 to December, 1993 (epidemic period) to determine risk factors for HIV secronconversion. Haemodialysis and medical records were reviewed, dyalisis centre staff and surviving patients were interviewed, and dialysis practices were observed. Stored sera from all dialysis centre patients were tested for HIV antibody.12 (52%) of 23 patients tested positive for HIV antibody by enzyme immunoassay and western blot during the epidemic period. Of the 23 tested, 9 (39%) converted from HIV antibody negative to positive (seroconverters) and 10 (44%) remained HIV negative (seronegatives). The HIV seroconversion rate was higher among patients dialysed at the centre while a new patient, who was HIV seropositive was dyalised there (90% vs 0%; p<0·01), or when the dialysis centre reprocessed access neddles, dialysers, and bloddines (60% vs 0%).While 2 of 9 HIV seroconverters had had sex with prostitutes, none had received unscreened blood products of had other HIV risk factors. No surgical or dental procedures were assocuated with HIV seroconversion. Dialysers were reprocessed separately with 5% formaldehyde and were labelled for use on the same patient. Access needles were reprocessed by soaking them in a common container with a low-level disinfectant, benzalkonium chloride; 4 pairs of needles were placed in one pan creating the potential for cross- contamination or use of one patient's needles on another patient.HIV transmission at the dialysis centre was confirmed. Improperly reprocessed patient-care equipment, most probably access needles, is the likely mechanism of transmission. This outbreak was discovered by accident and similar transmission may be iccurring in many other countries where low-level disinfectants are used to sterilise critical patient-care equipment.