PURPOSE: To develop a reproducible surgical technique for the induction of choroidal neovascularization (CNV) in the subretinal space of porcine eyes and to analyse the resulting CNV clinically and histologically. METHODS: Two different modifications of a surgical technique previously described were compared with the original method. In ten porcine eyes retinal pigment epithelial (RPE) cells were removed using a silicone tipped cannula, in ten porcine eyes Bruch's membrane was perforated once with a retinal perforator without prior RPE removal and in ten eyes RPE removal was followed by a single perforation of Bruch's membrane. Fifteen of the eyes, five from each group, were enucleated 30 minutes after surgery, while the remaining eyes were enucleated after 14 days. Prior to enucleation, at day 14, fundus photographs and fluorescein angiograms were obtained. Eyes were examined by light microscopy and by immunohistochemical staining. In addition to these 30 eyes, two eyes underwent surgery with the purpose of subsequent scanning electron microscopic (SEM) examination. RESULTS: In eyes enucleated immediately after surgery neuroretinas overlying the induced lesions were intact without apparent atrophy of cells regardless of the surgical technique applied. The process of RPE removal was found to induce breaks in Bruch's membrane and both the size and the number of breaks varied between eyes. CNV membranes were identified in 15 of 15 eyes enucleated after 14 days. CNV membranes induced by perforation of Bruch's membrane without prior RPE removal were significantly thicker than membranes from eyes undergoing both RPE removal and Bruch's perforation (p = 0.03) and also thicker than membranes from eyes with only RPE-removal (p < 0.01). CNV membranes from eyes with perforation of Bruch's membrane without prior RPE removal had a higher cellular content and were more richly vascularized and also exhibited the highest propensity to leak in fluorescense angiograms. CONCLUSION: All three surgical techniques were capable of inducing CNV, but the one applying perforation of Bruch's membrane without RPE removal was easier to reproduce and involved fewer variables than the techniques utilizing RPE removal. The presence of RPE cells seems to affect both the morphology and cellular composition of induced CNV.