Methicillin-resistant Staphylococcus aureus (MRSA) has presented special problems in intensive care units (ICUs) because of the difficulties in implementing infection control measures. The prevalence and rate of acquisition of MRSA were studied over thirty months in a nine-bed ICU. Nasal and groin swabs were taken on admission and then weekly, and other cultures as clinically indicated. Of 1361 admissions 119 were MRSA-positive on arrival. 21 cases had been identified before admission and the remainder were detected by screening; in 57 the positive result was known only after discharge. Of the 1242 admissions initially negative 68 acquired MRSA while in the ICU. The ICU had no known MRSA-positive patients on 185 (20.3%) of 914 days, the longest sequence being 17 days. Positive patients occupied 1387 (16.9%) of the 8226 available bed days. Length of stay predicted the risk of acquiring MRSA. Estimated from patients who completed each weekly screening cycle, the risk was 7.5% per week in the first week and 20.3% per week thereafter. The risk was not influenced by initial APACHE II score, the use of haemofiltration, or the number of MRSA-positive patients in the unit. The data suggest that a further 38 of those discharged between weekly screenings acquired MRSA, giving an incidence of 8.5%. MRSA was grown from blood in 17 patients, and from sputum in 53 (ICU-acquired in 18% and 47%). This study suggests that nearly 10% of admissions to a general ICU will be MRSA-positive, of whom only half will be identified before discharge. With standard prevention the risk of previously negative patients acquiring MRSA approximates to 1% per day in the first week and 3% per day thereafter, with nearly one-fifth progressing to bacteraemia; one-half will have MRSA in sputum. Patients with longer stays constitute a high-risk minority for whom additional measures such as decontamination with oropharyngeal and enteral vancomycin should be considered.