Selective decontamination of the digestive tract (SDD) with oral nystatin was evaluated as a measure to control an outbreak of Candida infection in a neonatal intensive care unit (NICU). Seventy-six out of 106 neonates who carried Candida spp. received the main study manoeuvre (the application of oral nystatin in the throat and stomach) during the 12-month open trial. One third of the neonates weighed < 1500 g whilst about half were being ventilated. The mean stay was 33.2 d (SD +/- 46.9). Two cases with candidaemia within a fortnight were associated with a yeast carriage rate in the NICU of about 50%; more than 80% of the isolates were Candida parapsilosis. During the implementation period there were four new neonates with fungaemia caused by C. parapsilosis. Once the carriage rate dropped below 5% (P < 0.001), no new cases of systemic infection with the outbreak strain were recognized in the following 8 months. It took 3.5 months to control the outbreak. The observation that all other clinical diagnostic samples were free from Candida suggests that translocation from throat or gut into the systemic circulation occurred. SDD with oral nystatin was effective in reducing the yeast carriage index (mean index 1.93, before SDD; 0.45, after SDD; P < 0.001). A significant reduction of carriage, both in rates and indices, is thought to have contributed to the control of this candida outbreak.