The use of the Clinitest urine glucose test as an indirect way to monitor and control blood glucose levels in critically ill patients was studied. Part 1 of the study was a review of 16 patients records to determine the correlation between 348 paired Clinitest readings and blood glucose levels. Part 2 was a study of six patients from whom a total of 83 arterial blood samples were obtained, each followed 15 minutes later by a urine sample that was tested quantitatively by autoanalyzer and qualitatively by Clinitest. In Part 1, only 43% of the paired readings were within clinically acceptable levels. In Part 2, there was an 11% error rate between quantitative urine glucose values and the corresponding Clinitest results, and the relationship between arterial and urine glucose concentrations was unpredictably variable. The study indicates that both Clinitest error and a variable relationship between arterial and urine glucose concentrations make the use of Clinitest as the sole monitor and basis for controlling blood glucose levels in critically ill patients as unsatisfactory and potentially dangerous procedure.