To assess variations in hospital length of stay for community acquired pneumonia patients adjusted for their severity of illness using a pneumonia specific model as described by Fine et.al in the N Engl J Med 1997;336;243-50. METHODS: All patients admitted to University Hospital, San Antonio from January-December 1996 with a primary ICD-9CM diagnosis of 486 (pneumonia with no organism specified) were evaluated retrospectively via chart review. Patient demographics data, clinical findings, hospital resource utilization such as length of stay, length of antibiotic therapy and hospital mortality were collected. Patients were stratified into 5 risk classes. RESULTS: Total of 187 patients were evaluated. There were (36,60,42,40,9) patients in risk classes (I, II, III, IV, V) . Hospital length of stay for patients in classes (I, II, III, IV, V) were in order LOS£3 days ( 5, 13, 12, 3, 2) , LOS 4-7 days (24, 39, 20, 18, 1) and LOS >7 days (7, 8, 10, 19, 6) respectively. Average length of hospital stay were (6.6, 5.1, 6.2, 8.1, 9.6) days in class I-V respectively, Anova P<0.0026. CONCLUSION: Variations exist in hospital LOS among the five groups. 81% (78 patients in I and II combined) were treated > 3 days and 71% (30 patients in class III) were treated > 3 days. These low to moderate risk classes (I, II, III) could be a target for reduced LOS, quality assurance and cost effectiveness program. These differences in LOS could be attributed to other unmeasured patient, physician or hospital related factor and its process of care.