Abstract Groups of 12 male and 24 female 5-wk-old Charles River CD ®(SD) BR rats (F 0 were fed a sucrose-containing ground cereal-based diet in which 0, 2.5, 5.0 and 10.0% (w/w) sorbitol was included at the expense of sucrose. The rats were first mated after 14 wk on the diet. F 1a litters were born 19 wk after the start of the study and F 1b litters at wk 30. Groups of 12 male and 24 female F 1b rats were first mated when 18 wk old. They gave rise to F 2a litters after 3 wk and to F 2b litters 10 wk later. Likewise, groups of 12 male and 24 female F 2b rats were first mated when 18 wk old, producing F 3a and F 3b litters 3 wk and 10 wk later, respectively. F 0 rats were killed 33 wk after the start of the study, F 1a in wk 22, F 1b in wk 68, F 2a in wk 57, F 2b in wk 92 and F 3a in wk 96. Apart from slight reductions in food consumption in sorbitol-fed F 1b males and in body-weight gain in sorbitol-fed F 0, F 1b and F 2b rats of both sexes, treatment was associated with no clinically observed effects. There were no deaths attributable to treatment and no adverse effects on mating performance or pregnancy rates in the parent animals of any generation. Treatment was associated with no consistent adverse effect on any measure of reproductive performance or behaviour during gestation or lactation. No abnormal pups were observed in any generation. Not unexpectedly, caecal enlargement was consistently observed at necropsy of sorbitol-treated rats of all generations and significant rises in serum calcium were observed in F 0 males and females exposed to 10% sorbitol and in F 1b males exposed to either 5 or 10% sorbitol. Differences between treated and control F 3a rats in respect of T 3 and TSH levels were probably spurious as they followed no consistent pattern. Similarly, between-group variations in gonadal weight were considered to have no toxicological significance because they lacked consistency and were not accompanied by any histologically-evident changes. Microscopic examination of lesions from F 1a and F 2a animals, of gonads from F 1b and F 2b and of selected tissues from the F 3a generation revealed no changes of toxicological significance. A reduced incidence of hepatocytic swelling in 10% sorbitol-treated F 3a females was thought to reflect no more than the slight difference in nutritional status between these animals and the controls. No abnormalities of the adrenal medulla were seen grossly in any generation or microscopically in the high-dose and control F 3a rats. It is concluded that sorbitol administered in the diet to three successive generations of rats at levels up to 10% had no adverse effect on growth or reproductive performance in either sex.