Mupirocin has been widely used for the clearance of nasal methicillin-resistant Staphylococcus aureus (MRSA) carriage during outbreaks, but no placebo-controlled trial has evaluated its value for eradicating MRSA carriage at multiple body sites in settings where MRSA is not epidemic. In a 1,500-bed teaching hospital with endemic MRSA, 102 patients colonized with MRSA were randomized into a double-blind, placebo-controlled trial and treated with either mupirocin (group M) or placebo (group P) applied to the anterior nares for 5 days; both groups used chlorhexidine soap for body washing. Follow-up screening, susceptibility testing, and genotyping were performed to evaluate treatment success, mupirocin or chlorhexidine resistance, and exogenous recolonization. At baseline, MRSA carriage was 60% in the nares, 38% in the groin, and 62% in other sites (skin lesions, urine). The MRSA eradication rate (all body sites) was 25% in group M (12 of 48 patients), compared to 18% in group P (9 of 50 patients; relative risk [RR], 0.72; 95% confidence interval [CI95], 0.33 to 1.55). At the end of follow-up, 44% of patients (19 of 43) were free of nasal MRSA in group M, compared to 23% (11 of 44) in group P (RR, 0.57; CI95, 0.31 to 1.04). Ten patients developed MRSA infections (three in group M and seven in group P). One mupirocin treatment failure was due to exogenous MRSA recolonization. No MRSA isolate showed chlorhexidine resistance or high-level mupirocin resistance; however, we observed an association (P = 0.003) between low-level mupirocin resistance at study entry (prevalence, 23%) and subsequent treatment failure in both study arms. These results suggest that nasal mupirocin is only marginally effective in the eradication of multisite MRSA carriage in a setting where MRSA is endemic.