We investigated the short-term effects of moderate increments or decrements of protein intake on albumin excretion rates of type I diabetic patients. Ten normotensive patients with either normal albumin excretion rates (< 20 micrograms/min, group I, n = 5) or persistent mu-albuminuria (20-200 micrograms/min, group II, n = 5) were fed successively three test diets providing different protein intakes. Each patient was randomly allocated to 3-wk sequences of the following diets: low-protein diet (LPD; median 0.84, range 0.76-0.94 g.kg-1.day-1), medium-protein diet (MPD; median 1.33, range 0.98-2.00 g.kg-1.day-1), and high-protein diet (HPD; median 2.05, range 1.54-2.61 g.kg-1.day-1). The three diets were isoenergetic and isoglucidic. In group I patients, no consistent change was found in mu-albuminuria. In group II patients, LPD resulted in a reduction in mu-albuminuria compared with MPD and HPD. Changes in albumin excretion rates were positively correlated to relative changes in protein intake. This suggests that moderately protein-restricted diets can reduce mu-albuminuria in diabetic patients suffering from incipient nephropathy, the degree of reduction depending on the degree of restriction. Because of poor patient compliance with protein intakes < 0.8 g.kg-1.day-1, we conclude that moderately rather than severely protein-restricted diets should be recommended for long-term prescriptions.