Creatinine measurement by alkaline picrate reagents is subject to positive interference by acetoacetate. Enzymatic reagents avoid this interference and have been adapted to instruments such as the Ektachem-400 (Kodak). By documenting the discrepancy between alkaline picrate and Ektachem determinations for creatinine, the authors prospectively identified ketotic patients in whom the presence of ketones was responsible for a significant creatinine error. During their three-month survey, they identified 50 such ketotic inpatients. Those admitted to the medicine service represented almost 5% of all medicine admissions over this time. Of the total specimens, the mean discrepancy was 14 +/- 8 mg/L with a range of 4-44 mg/L. The greater the ketosis, the greater the discrepancy. Two-thirds of the samples were normal on the Ektachem but greater than normal by picrate methods. In addition to diabetes or ethanol abuse, 17% of the ketotic patients had severe or terminal illness that was generally associated with malnutrition.