OBJECTIVE: To estimate the number of patients with endophtalmitis hospitalized in France, its average cost and hospital financing consequences. METHODS: The French PMSI (Programme de Médicalisation des Systèmes d'Information) 2003 database was analysed. Patients with the following principal diagnoses were extracted: “purulent endophthalmitis”, “other endophthalmitis”, and “endophthalmitis associated with another disease”. Two mean lengths of stay (LoS) were compared: the overall patient sample and the DRG-weighted estimate. Endophthalmitis costs were estimated as the sum of weighted average DRG costs plus length of stay multiplied by hospital variable costs. RESULTS: 1681 endophthalmitis cases were reported (age 66.7, 50% male). 1449 cases (86.2%) were in the public sector and 2.8% were ward transfers. A total of 221 (13.1%) had a vitrectomy, the most common endophtalmitis-related surgery. A total of 85 (5.1%) required a ward transfer and 4 (0.2%) died. The patient LoS was 7.6 days on average while the DRG weighted LoS was 4.3 days. Thus, the PMSI underestimates endophtalmitis LoS by 3.3 days. Mean costs exceeded the weighted DRG cost by €1396 for the public and €813 for the private hospital. Overall endophtalmitis costs were euro 4125. Using DRG 60 (Severe acute ocular infections) as a proxy of endophthalmitis would underestimate dramatically (>45%) its cost. Endophthalmitis expenses were estimated at €6,509,000 for the French Health Care System. CONCLUSION: Using DRG 60 as a proxy of endophthalmitis in health economics evaluation to estimate the hospital true cost is inappropriate The PMSI clustering algorithm underestimates the budget allocation required to treat endophtalmitis in the public and private sectors. The PMSI (reporting exhaustively all hospitalisations) could be used to better capture the yearly endophthalmitis incidence rate.