We determined risk factors associated with persistent carriage of methicillin-resistant Staphylococcus aureus (MRSA) among 102 patients enrolled in a double-blind, placebo-controlled trial of nasally administered mupirocin ointment. MRSA decolonization was unsuccessful in 77 (79%) of 98 patients who met the criteria for evaluation. By univariate analysis, 4 variables were found to be associated with persistent MRSA colonization (P < .1 for all 4): absence of mupirocin treatment, previous fluoroquinolone therapy, > or = 2 MRSA-positive body sites, and low-level mupirocin resistance. After multivariable Cox proportional hazards modeling, the presence of > or = 2 positive body sites (adjusted hazard ratio [AHR], 1.7; 95% confidence interval [CI], 1.0-2.9) and previous receipt of a fluoroquinolone (AHR, 1.8; 95% CI, 1.0-3.3) were independently associated with MRSA persistence, whereas nasal mupirocin tended to confer protection (AHR, 0.6; 95% CI, 0.4-1.0). Low-level mupirocin resistance was observed in 9 genotypically different MRSA strains and was not independently associated with chronic MRSA carriage (AHR, 1.5; 95% CI, 0.9-2.5). Our findings suggest that multisite MRSA carriage and previous receipt of a fluoroquinolone are independent risk factors for persistent MRSA colonization.