Little is known about the amount of cross-transmission, the risk factors for infection, and the relative effectiveness of infection control procedures when methicillin-resistant Staphylococcus aureus (MRSA) infection occurs at highly endemic levels in intensive care units. A cohort study was done to identify exposures associated with cases that likely were the result of cross-transmission (i.e., occurring in clusters and with indistinguishable MRSA macrorestriction profiles). Fitting a simple stochastic model to the ascertained data allowed prediction of the effectiveness of infection control measures. Exposure to relative staff deficit (adjusted rate ratio, 1.05 independent; 95% confidence interval, 1.02-1.09) was the only factor significantly associated with potential transmission (P =.001). It was predicted that a 12% improvement in adherence to hand-hygiene policies might have compensated for staff shortage and prevented transmission during periods of overcrowding, shared care, and high workload but that this would be hard to achieve.