We investigated the effects of prolonged treatment of diabetic rats with curcumin-supplemented yoghurt on the physiological and biochemical changes associated with diabetes mellitus. An established metabolic cage model was used to assess these changes in three groups of streptozotocin-diabetic rats which had been administered, by gavage, curcumin blended into yoghurt in the doses of 30, 60 and 90 mg/kg body weight (BW) per d (groups DC30, DC60, DC90) for 31 d. One group of non-diabetic rats was also treated with 90 mg/kg BW per d curcumin (NDC90). Three control groups of diabetic animals received water (DW), yoghurt (DY) and insulin at 27·78 μmol/d by subcutaneous injection (DI). Also, two groups of non-diabetic animals received water (NDW) and yoghurt (NDY). Groups DI and DC90 exhibited significant falls, relative to DW and DY, in food and water intake, urine volume, glycaemia, urinary urea and glucose, proteinuria, serum TAG and activities of aspartate and alanine aminotransferases, and higher hepatic glycogen and BW. These improvements were greater in DI than in DC90. No difference was observed in the serum levels of total cholesterol or HDL-cholesterol, or in the masses of adipose and muscular tissues, between DC90 and DW or DY. Moreover, the improvements in curcumin-treated rats, relative to DW and DY, were significant and dose-dependent. The NDC90 group also showed no difference from the NDW or NDY groups, in any of the markers for diabetes. In conclusion, curcumin mixed into yoghurt at the highest dose tested exhibited anti-diabetic activity, improving significantly most of the markers assessed in this study.