Groups of 114 (control) or 66 (treated) rats of each sex were fed diets providing 0 (control), 100, 400 or 1200 mg carmoisine/kg body weight/day for 9 wk. Within each group the animals were mated monogamously. Treatment continued uninterrupted until the young were randomly selected (where possible one/sex/litter) from each of the litters to provide groups of 90 (control) and 54 (treated) rats of each sex. These received the same treatment as their parents for up to 110 wk for females or 115 wk for males. Apart from coloration of the fur, urine and faeces, treated rats did not differ in appearance or behaviour from the controls. Mortality was not affected by treatment. High-dose groups had reduced body-weight gain compared with that of the controls, despite slightly higher food intakes. Increased water intakes in the same animals accompanied a tendency to excrete larger volumes of urine. Haematological investigations at months 3, 6, 9, 12, 18 and 24, and on all survivors at the end of the study showed no treatment-related effects. Urine studies on 20 rats/sex/group at months 3, 6, 12, 18 and 24 showed no consistent treatment-related changes. Analyses of serum collected at the end of the study demonstrated low glucose concentrations in both sexes of the high-dose group and in intermediate-dose females. A few high-dose males had bladder hyperplasia while some high-dose females had increased incidences of adrenal blood/fibrin cysts or internal hyperplasia/medial hypertrophy of the pancreatic blood vessels. Tumours occurred with a similar distribution and incidence in all groups apart from an increased incidence of adrenal phaeochromocytoma in high-dose males. The incidence seen was well within the background incidence for this relatively common tumour in the same strain of rat under similar conditions. It is concluded that carmoisine is not carcinogenic and that the no-untoward-effect level in this study was 400 mg carmoisine/kg body weight/day.