To determine the risk of nephrotoxicity induced by the infusion of radiographic contrast material, we undertook a prospective study of consecutive patients undergoing radiographic procedures with intravascular contrast material. There were three study groups: patients with diabetes mellitus and normal renal function (n = 85), patients with preexisting renal insufficiency (serum creatinine level, greater than or equal to 150 mumol per liter) without diabetes (n = 101), and patients with both diabetes and renal insufficiency (n = 34). The control group consisted of patients undergoing CT scanning or abdominal imaging procedures without the infusion of contrast material who had diabetes mellitus (n = 59), preexisting renal insufficiency (n = 145), or both (n = 64). Clinically important acute renal failure (defined as an increase of greater than 50 percent in the serum creatinine level) attributable to the contrast material did not occur in nondiabetic patients with preexisting renal insufficiency or in diabetics with normal renal function. The incidence of clinically important contrast-induced renal failure among the diabetic patients with preexisting renal insufficiency was 8.8 percent (95 percent confidence interval, 1.9 to 23.7 percent), as compared with 1.6 percent for the controls. The incidence of acute renal insufficiency, more broadly defined as an increase of greater than 25 percent in the serum creatinine level after the infusion of contrast material, was 11.8 percent among all patients with preexisting renal insufficiency. After the exclusion of patients whose acute renal insufficiency could be attributed to other causes, the incidence was 7.0 percent (95 percent confidence interval, 3.2 to 12.8 percent), as compared with 1.5 percent in the control group. The risk of acute renal insufficiency attributable to the contrast material was therefore 5.5 percent, and the relative risk associated with the infusion of contrast material was 4.7. These rates were similar whether the osmolarity of the contrast material was high or low. We conclude that there is little risk of clinically important nephrotoxicity attributable to contrast material for patients with diabetes and normal renal function or for nondiabetic patients with preexisting renal insufficiency. The risk for those with both diabetes and preexisting renal insufficiency is about 9 percent, which is lower than previously reported.