The incidence of potentially erroneous tests for urinary glucose was studied. Each urine specimen collected was tested with Clinitest, Clinistix, Diastix, and Tes-Tape before and after addition of glucose to produce a urine glucose concentration of 1/2 per cent. There was a 23 per cent incidence of falsely high (>1/ 2 per cent) and a 33 per cent incidence of falsely low (<1/2 per cent) results in the 513 specimens examined. Underreading, frequently caused by urinary metabolites of common medications, was seen only with Clinistix and Diastix. Overreading had a distinct relationship to dilute urine and was most frequent with Clinitest (65 per cent). The patient's habits (medications taken and water ingestion) should be considered carefully when a urine testing method is selected. Periodic plasma glucose measurements are recommended to confirm impressions obtained from urine tests for glucose.