We conducted this study to determine the associated factors and outcomes of community-acquired primary bacteremia (PB), to describe the most frequently isolated microorganisms, the antibiotic resistance pattern, and to guide the most appropriate antibiotic treatment. A total of 1,640 community-acquired bacteremia cases in nonneutropenic adults were consecutively enrolled from January 2003 to May 2006 and prospectively followed up. Nonconditional logistic regression methods were used with PB and death as dependent variables. Secondary bacteremia (SB) was present in 1,440 patients, and 200 (12%) cases were PB. The independent factors associated with PB were male sex (OR 1.69, 95%CI 1.27-2.25, P = 0.001) and an ultimately or rapidly fatal prognosis for an underlying disease (OR 2.48, 95%CI 1.84-3.34, P = 0.001). The most frequently isolated microorganisms in PB were E. coli and other enterobacteria (26 and 22%, respectively) and S. aureus (15%). There were 28 and 37% ciprofloxacin-resistant E. coli strains in SB and PB, respectively (P = 0.2). Mortality was significantly higher in PB cases (13 vs 8%, P = 0.04). The independent factors associated with mortality in PB were ultimately or rapidly fatal prognosis of underlying disease (OR 2.1, 95%CI 1.41-3.13, P = 0.001), lack of fever at the moment of bacteremia (OR 2.38, 95%CI 1.18-4.76, P = 0.02) and incorrect empirical antibiotic therapy (OR 2.01, 95%CI 1.22-3.33, P = 0.006). The initial empiric antibiotic treatment was more frequently incorrect in PB than in SB, and this was a predictive factor for mortality in PB. The resistance pattern of E. coli, other enterobacteria and S. aureus in every setting should guide the most appropriate empirical treatment for PB.