Nosocomial infection rates are frequently calculated by dividing the number of infections acquired during a month by the number of patients discharged during that month. This method does not allow for prediction of future events or comparability between locations for statistical analysis. A more appropriate rate is the number of infections divided by patient-days spent in the hospital. This rate avoids the problems associated with the commonly used method. On the basis of the rate of infections per patient-day, it is possible to test for interhospital and intrahospital differences as well as for differences over time.