To clarify the mechanism by which renal potassium (K) excretion is reduced in children with insulin-dependent diabetes mellitus, we studied two groups of patients: (A) at diagnosis and (B) after at least 1 year of follow-up. Group A (15 children) was studied twice: on the day of admission and after 1 month of insulin therapy. On admission, urinary K excretion, fractional K excretion, and transtubular K concentration gradient (TTKG) were significantly decreased, but became normal after extended insulin therapy. TTKG was inversely correlated with blood glucose (P < 0.001) and hemoglobin A1c (HbA1c, P < 0.001). Group B (73 children with a mean follow-up of 54 +/- 36 months) was subdivided according to the TTKG: 30 patients had a low TTKG < 4.0 (median 3.2) and 43 patients had a normal TTKG > or = 4.0 (median 5.2). Patients had a low TTKG and those with a normal TTKG had an identical duration of follow-up and similar values for plasma renin activity, aldosterone concentration, calciuria, magnesiuria, albumin excretion rate, and creatinine clearance. However, those with a low TTKG had significantly higher blood HbA1c levels, urine volume, and glucosuria. Logistic regression analysis showed that the only independent variables predicting a low TTKG were blood HbA1c and glucosuria (P < 0.001). These data confirm that a reduced renal K excretion is a characteristic feature of diabetic children; this is reversible with appropriate insulin therapy, largely depends on the metabolic control of the disease, and, specifically, on the degree of hyperglycemia and/or glucosuria.