The aims of the Australian Corneal Graft Registry are to collect and collate statistical information on the practice of corneal transplantation around Australia, to identify risk factors for corneal graft failure, and to provide information on graft and visual outcome. The current report encompasses analyses performed on 3608 corneal grafts (96% penetrating and 4% lamellar) entered into the Registry between May 1985 and July 1991. Sixty-four per cent of grafts have undergone one or more rounds of follow-up by the 189 contributing surgeons and 110 additional referring practitioners: five-year Kaplan-Meier graft survival for penetrating and lamellar grafts is 72% and 84%, respectively. The main indications for penetrating keratoplasty were keratoconus (31%), bullous keratopathy (25%), history of failed previous graft (14%), corneal scars and opacities (11%), and corneal dystrophies (7%). The most common reasons listed for failure of penetrating grafts were rejection (33%), glaucoma (11%), non-viral infections (10%), endothelial cell failure (8%) and herpetic infection (7%). In 19% of cases, the reason for graft failure was unclear. The main indications for lamellar keratoplasty were pterygium (32%), thinning, necrosis or ulceration from old beta-radiation therapy for pterygium (17%), and scleral ulcers, necrosis, ectasia, perforations or melts (29%). The most common reasons for the failure of lamellar grafts were corneal melting (43%) and sloughing of the graft (29%). Among the factors that influenced the survival of penetrating corneal grafts to a significant extent (P < 0.05) in univariate analysis were: the centre effect, indication for graft, graft number, a history of pregnancy or blood transfusion, inflammation before or at the time of graft, corneal vascularisation at the time of graft, a history of raised intraocular pressure, the donor cornea procurement source, the death to donor cornea enucleation time, graft size and large degrees of oversizing, lens status and the type of intraocular lens in situ. In the postoperative period, risk factors for failure included early removal of graft sutures, neovascularisation of the graft, herpetic recurrences in the graft and the occurrence of rejection episodes. The variables that best predicted penetrating corneal graft failure in Cox proportional hazards regression analysis were aphakia or the presence of an anterior chamber of iris-clip intraocular lens, very small or very large grafts, a history of previous ipsilateral graft, an indication for graft that was neither keratoconus nor any of the corneal dystrophies, inflammation at the time of graft, and a postoperative rise in intraocular pressure.(ABSTRACT TRUNCATED AT 400 WORDS)