Practice guidelines recommend at least 14 days of antibiotic therapy for uncomplicated Staphylococcus aureus bacteremia (SAB). However, these recommendations have not been formally evaluated in clinical studies. To evaluate the duration of therapy for uncomplicated SAB, we analyzed data from our prospective cohort of patients with SAB. A prospective observational cohort study was performed in patients with SAB at a tertiary-care hospital in Korea between August 2008 and September 2010. All adult patients with SAB were prospectively enrolled and observed over a 12-week period. Uncomplicated SAB was defined as follows: negative results of follow-up blood cultures at 2 to 4 days, defervescence within 72 h of therapy, no evidence of metastatic infection, and catheter-related bloodstream infection or primary bacteremia without evidence of endocarditis on echocardiography. Of 483 patients with SAB, 111 met the study criteria for uncomplicated SAB. Fifty-three (47.7%) had methicillin-resistant SAB. When short-course therapy (<14 days) and intermediate-course therapy (≥14 days) were compared, the treatment failure rates (10/38 [26.3%] versus 16/73 [21.9%]) and crude mortality (7/38 [18.4%] versus 16/73 [21.9%]) did not differ significantly between the two groups. However, short-course therapy was significantly associated with relapse (3/38 [7.9%] versus 0/73; P = 0.036). In multivariate analysis, primary bacteremia was associated with a trend toward increased treatment failure (P = 0.06). Therefore, in the treatment of uncomplicated SAB, it seems reasonable to consider at least 14 days of antibiotic therapy to prevent relapse, as practice guidelines recommend. Because of its poor prognosis, primary bacteremia, even with a low risk of complication, should not be treated with short-course therapy.