We studied 11 healthy rabbits and 15 rabbits with experimentally induced Candida endophthalmitis. Using a microbiological procedure, ketoconazole levels were determined 1, 2, 4, 8 and 24 h after its oral administration. Ketoconazole penetration was average in the aqueous humour, significant in tears and poor in the vitreous. Vitreous penetration was not detectable in healthy rabbits or in rabbits with chorioretinits exhibiting little or no vitreous reaction. Such penetration was only significant in rabbits exhibiting massive vitreous exudation; however, in such rabbits, penetration was not always observed (only in 2 out of 6 animals). Given these results, the treatment of Candida endophthalmitis in humans using ketoconazole would appear to be of questionable value.