We performed an observational analysis of prospectively collected data on 533 nonseverely immunosuppressed adult patients who were hospitalized for community-acquired pneumonia on the basis of conventional admission criteria. For this population, we correlated etiology, reasons for admission, and outcomes using the Pneumonia Severity Index (PSI), to identify major discrepancies between the PSI risk class and the conventional criteria for deciding the site of care. PSI classes and corresponding mortality rates were as follows: class I, 51 patients (0%); class II, 62 (2%); class III, 117 (3%); class IV, 198 (10%); and class V, 105 (29%). We identified significant discrepancies between both methods. Overall, 230 patients (40%) who were hospitalized according to conventional criteria were assigned to low-risk classes. Of these 230 patients, 137 (60%) needed supplementary oxygen or had pleural complications; for the remaining patients, there were no irrefutable reasons for admission. This latter group deserves prospective evaluation in randomized studies that compare ambulatory and in-hospital management.