The continued spread of penicillin-resistant pneumococci raises therapeutic concerns. Optimal therapy for resistant infections is unknown and it is not clear whether the efficacy of penicillin or equally active beta-lactam agents is compromised in non-meningeal-resistant infections. A prospective nonintervention study was undertaken to compare the clinical response in penicillin-resistant vs. penicillin-susceptible bacteremic pneumococcal infections, excluding meningitis. Of 108 children enrolled, 35 (32%) had penicillin-resistant (one highly resistant) isolates. Seventy-eight children had pneumonia, 21 had occult bacteremia (sepsis) and 9 had peritonitis. Children with resistant infections were more likely to have underlying disorders, especially human immunodeficiency virus infection, and to have received antimicrobial therapy in the previous month. After 48 hours of therapy 64% of penicillin-susceptible infections showed improvement vs. 60% of penicillin-resistant infections (odds ratio, 1.2; 95% confidence intervals, 0.5 to 3.0). In children with pneumonia treated with ampicillin or an equivalent beta-lactam agent, 93% with penicillin-susceptible infections had improved by Day 7 of therapy compared with 88% with resistant infections (odds ratio, 1.9; 95% confidence interval 0.3 to 15.9). The durations of respiratory distress, fever and oxygen requirement were similar in penicillin-susceptible and -resistant infections. These results suggest that intermediate penicillin resistance is of little significance in pneumococcal pneumonia or sepsis and that standard beta-lactam therapy is still highly effective. Further studies of highly penicillin-resistant infections are necessary.