Rabbit corneas were incubated, over 4 h in vitro, with the corneal epithelial surface exposed to various solutions to assess their utility as incubation solutions for physiological, pharmaceutical or toxicological studies. The corneal endothelium was perfused with a 35 mm bicarbonate–mixed salts solution equilibrated at 36°C. Corneal thickness, corneal hydration or epithelial cell appearance (as assessed by scanning electron microscopy) were found to be similar to in vivo if a 35 mm bicarbonate, mixed salts solution (equilibrated with 5% CO2–air) was used for the epithelium. Some swelling (14 μmh−1), increased hydration and minor cell exfoliation were seen if this 35 mm bicarbonate solution was equilibrated with 5% CO2–95% O2 (hyperoxia). Solutions with only 5 mm bicarbonate (0.5% CO2–air) produced rapid swelling, large increases in hydration and marked cellular damage. Slightly hypertonic (310 mOsm kg−1) solutions containing 5 mm bicarbonate caused some swelling at 15 μm h−1, small increases in hydration and some cell damage but the swelling and cellular damage were further reduced by making the solution slightly more hypertonic (325 mOsm kg−1) by addition of NaCl and KCl. Saline (NaCl 0.9% or 0.97%) or phosphate-buffered saline (PBS) (300 mOsm kg−1) produced swelling at 21–28 μm h−1, 30% increases in hydration and almost total destruction of the superficial cell layers. These studies confirm in vivo experiments that saline (and also buffered saline solutions) are rather toxic to the corneal epithelium and thus should not be used as epithelial incubation solutions. Even when using mixed salts solutions and even with bicarbonate present, small differences in composition can have marked effects on corneal thickness, hydration or cell appearance. Hyperoxic solutions appear to be mildly cytotoxic compared with normoxic solutions.