BACKGROUND: The aim of our study was to develop possible criteria for the choice of vitrectomy in the treatment of endophthalmitis. MATERIALS AND METHODS: Ninety-two patients with endophthalmitis who were seen between 1980 and 1993 were reviewed retrospectively and analysed regarding pathogenesis, bacteriological aetiology, as well as pre- and postoperative visual acuity comparing intravenous (i.v.) therapy alone to a combination with vitrectomy in 3 pretherapeutic visual acuity classes (class I < 0.02, class II 0.02- < 0.1 and class III > or = 0.1). RESULTS: Endophthalmitis most commonly occurred after cataract extraction (39%). The best prognosis regarding final visual acuity was found with Staph. epidermidis infection, the worst with Streptococci. Postoperative visual acuity in class I was significantly better after vitrectomy with 0.18 +/- 0.06 (Standard Error = SE) compared to i.v. therapy alone (0.06 +/- 0.03 SE) (p < 0.05 unpaired Student t-Test). In class II, no such difference could be demonstrated. In class III, i.v. therapy yielded significantly better results (0.8 +/- 0.09 SE) than vitrectomy (0.53 +/- 0.14 SE), however only by comparing the final - but not the best - obtained posttherapeutic visual acuity. CONCLUSION: The more favourable results after vitrectomy in patients with very poor pretherapeutic vision suggest an important role of vitrectomy in the presence of extensive vitreous opacifications. The better outcome after i.v. therapy in patients with pretherapeutic vision > or = 0.1 could be influenced by the very short follow-up period which renders an evaluation of a potential loss of vision due to late complications after endophthalmitis impossible.