This perspective deals with prediction of overt diabetic nephropathy in patients with insulin-dependent diabetes mellitus (IDDM). The role of elevated urinary albumin excretion rate (microalbuminuria) in predicting diabetic nephropathy has been emphasized by new follow-up studies. Development of severe kidney impairment was seen in a large percentage of patients with microalbuminuria, but with more intensive care for diabetic patients, this percentage may be falling. Herein, I analyzed alternatives to microalbuminuria in predicting kidney disease in diabetes. 1) Parental predisposition to hypertension is not seen in all studies and therefore may not be a decisive factor, and it cannot be used in prediction of nephropathy. 2) Prediabetic blood pressure may predict nephropathy in certain non-insulin-dependent diabetic patients, but elevated blood pressure seems to develop after early microalbuminuria and is likely to be an aggravating factor in established microalbuminuria in IDDM patients. 3) At the clinical diagnosis of IDDM, diabetic nephropathy cannot be predicted. 4) Glycemic control is poor in normoalbuminuric patients with later development of microalbuminuria, and multiple glycosylated hemoglobin measurements are therefore important. 5) In diabetes, glomerular hyperfiltration is associated with late nephropathy, but it alone cannot be the decisive factor, because hyperfiltration in nondiabetic individuals does not produce kidney disease, according to new long-term follow-up studies. 6) Studies of glomerular structure and ultrastructure have not yet documented predictive values for overt nephropathy, but further studies are in progress. 7) Isolated blood pressure elevation without microabuminuria (probably representing essential hypertension in diabetes) has not been predictive. 8) It is clear that elevation of serum creatinine is a very late and insensitive parameter, occurring only with pronounced proteinuria. Creatinine clearance is a poor index for glomerular filtration rate. Therefore, it can be concluded that microalbuminuria currently is the most simple and sensitive parameter for detection of the patient at risk of kidney disease in diabetes. The abnormality probably reflects an early renal lesion rather than being a susceptibility factor. Long-term follow-up of microalbuminuria is recommended for early detection of diabetic kidney disease in IDDM patients, and proposals regarding intervention are discussed.