Summary. Male WLstar rats were maintained from the age of 4 weeks on one of four feeding programmes-ad lihititm feeding with chow with or without the addition of gluco.se to tlic drinking water, or meal feeding with restricted or unrestricted quantities of chow-to evaluate the effects of diet on body composition and metitl>oli.sm. In rats fed ad libitum on cliow, the percentage of water in the carcas.s decreased, the percentages of ash and protein rose, hut tliere was no change in the percentage of lipid during growth. The addition of glucose to the drinking water caused ti moderate elevation of the percentage nf lipid and a corresponding decrease iu the percentage of water in the carciiss during the first 2 weeks. Unrestricted meal feeding for only 2 horns per day caused the percentage of lipid in the carca.ss to incrpa.se during growth. This effect of meal feeding was aholi.slicd by taloric restriction. Botli tbe hody weight and the weight of the epididymal fat pad pro^ed to bi' unreliable indices of the percentage of lipid in tbp carcass.Tbe weight of the pancreas was directly propnrticinn! to body wt'ifiht. Tbe concentration of preformed insulin in the pancreas was elevated by unrestricted meal feeding and by suppleuienting the diet with gluco.se. Tbe rate of insulin .secretion in vitro by the pancreas lu the pre.sence of physiological (1 mg/ml) or high (5 mg/ml) concentrations of glucose was directly proportional to the concentration of prefonned insulin in the tissue. The rate of insulin secretion III titro of tissues from rats in tbe post-ahsorptivc phase was also directly pniportional to tbe total weigbt nf the rat. In nical-frd rats, which were in tbe late post-ab.sorptive phase when sacrificotl, the cituieutration of plasma insulin was directly proportional to hody vveigbt. the total insulin content of the pancreas and tbe rate of insulin secretion in vitro. Tbese findings suggest that f)otli dietary factors and total body mass detennine the supply of insulin and the hasal plasma iustilin concentration.The insulin re.sponses in vivo to intravenous glucose and the bypoglycaemic effect of intravenous insulin were found to be affected by tbe age of th(' rats and tbe metliod of feeding, hut in a highly variable manner.