The carcinogenicity of ammonia-process caramel, a food colouring, was examined in F344 rats. Caramel was dissolved in distilled water at levels of 0, 1 and 4% and groups of 50 male and 50 female rats were given 20-25 ml of one of these solutions/rat/day as their drinking water for 2 yr. There were no significant differences between the total incidences of tumours or mean survival times of control and experimental groups. A variety of tumours developed in all groups including the control group, and no dose-related effects were found either in the incidence or induction time of tumours in the various organs and tissues except in the pituitary gland of males, in which the incidence of tumours in males given 4% caramel solution was significantly higher than that in controls. Pituitary tumours are among the most common spontaneous tumours in ageing rats of this strain and have a variable incidence. In addition, almost all pituitary tumours detected in males given the 4% solution were microscopic tumours, and there was no significant difference between controls and treated groups in the incidence of hyperplasia or pre-neoplastic lesions in the pituitary gland. These results indicate that the significantly higher incidence of pituitary tumours in males given the 4% caramel solution was not related to caramel administration, but could be explained by the variability of the incidence of spontaneous pituitary tumours. Thus it is concluded that under these experimental conditions ammonia-process caramel was not carcinogenic in F344 rats.