A prospective microbiological surveillance (PMS) program was developed in a comprehensive hospital-wide effort for control of nosocomial methicillin-resistant Staphylococcus aureus (MRSA). This PMS program entailed: 1) active identification of colonized and infected patients; 2) application of a screening microbiologic system for MRSA; 3) isolation of colonized and infected patients; 4) antibiotic decolonization of MRSA; and 5) educational efforts. The PMS program was studied over three and one half years for its contribution to infection control of MRSA, early identification of nosocomial MRSA outbreaks, use of the highest yield surveillance culture sites, and cost effectiveness. Following initiation of the PMS program in December 1982, during an MRSA outbreak, the frequency of new MRSA cases declined from 14 to none by the end of a 3-month pilot study. The frequency of new MRSA cases stabilized at approximately 2 per month until October 1983, when the PMS system allowed prompt detection of a new outbreak of 11 cases. Following isolation and antibiotic decolonization, the frequency of cases again declined to 3 per month. A third outbreak in December 1985 again was promptly detected and controlled. Infection to colonization ratio decreased from a maximum of 1.5 during outbreaks to a minimum of 0.17 after outbreaks. Wounds and tracheostomy sites provided the greatest yield of detection of new cases of MRSA. During one 15-month period, 35 of the 43 new cases were detected initially at wounds and tracheostomy sites. No new MRSA cases were detected by a positive axillary or nares site alone. The estimated quarterly cost of outbreaks and infection paralleled the quarterly frequency of new MRSA cases. The cost of managing MRSA outbreaks and treating MRSA infections versus the cost of implementing the PMS program suggested that the PMS program may be cost effective. This prospective microbiological surveillance program may be applicable to other institutions for hospital-wide infection control of MRSA.