OBJECTIVES: The purpose of this study was to estimate the total and incremental costs associated with antibiotic treatments for community acquired pneumonia (CAP). METHODS: Persons over age 18 were identified in the Marketscan databases July-December 2004. We identified CAP episodes for patients with claims for ICD9 codes 481.XX-486.XX, plus the most frequent antibiotics prescribed as initial treatment for an episode. Logistic regression estimated a propensity score for each patient; which was the predicted probability of using telithromycin. Patients were then matched according to this probability. Then exponential conditional means models (ECM) were specified, controlling for variables that were still significantly different after the propensity score matching. These models allowed the incremental costs to be estimated for treatment of telithromycin relative to other antibiotics. RESULTS: The most common initial antibiotic treatments compared to telithromycin for CAP (n=187 CAP episodes) were amoxicillin, azithromycin, clarithromycin, levofloxacin, moxifloxacin, and a group of all other antibiotics. The mean length of a CAP episode was from 11.5 to 16.5 days. Mean total expenditures among the episodes was $897, with a range of $530 to $1175. Multivariate ECM models were fitted and showed significant incremental cost reductions per episode associated with telithromycin relative to: amoxicillin (-$283, p=0.002), azithromycin (-$686, p<0.001), levofloxacin (-$411, p<0.001), and moxifloxacin (-$484, p<0.001). The difference between telithromycin and clarithromycin (-$52) was not significant. CONCLUSION: The results of this study show that the costs differed among episodes of CAP by the initiating antibiotic. Use of propensity score matching and ECM regression controlled for intra-episode differences, so the incremental costs differences may be attributed to factors such as length of episode. The costs of CAP can be substantial and it is important to note that the initiating antibiotic may affect those costs.