Piperacillin as a single agent was compared in a prospective randomized trial with carboxypenicillin-aminoglycoside combinations in empirical therapy of serious bacterial infections. The difference in the clinical response rates with piperacillin (77% of 26 infection episodes) and combination therapy (75% of 24 infection episodes) were not statistically significant. Fewer adverse effects occurred in the piperacillin-treated group (42%) than in the combination-treated group (71%) (P = 0.0399 by Fisher's exact test), although neither nephrotoxicity nor hypokalemia alone was significantly less frequent in patients receiving piperacillin. However, the emergence of resistant organisms during therapy was more frequent among patients receiving piperacillin alone (42% of patients) than among patients receiving combination therapy (17% of patients) (P = 0.465 by Fisher's exact test). Moreover, emergence of resistance accounted for 5 of 9 patients with treatment failure, superinfection, or both when piperacillin was used as a single agent, compared with 2 of 10 similar patients in the combination group (P = 0.1299 by Fisher's exact test). The use of piperacillin as a single agent in the treatment of serious bacterial infections is not advocated, and the addition of an aminoglycoside to prevent emergence of resistance during empirical therapy of such infections is strongly recommended.