It has been suggested that a single random blood glucose measurement, timed in relation to food, can be used to determine those women needing a formal oral glucose tolerance test for the detection of gestational diabetes. One hundred and ten pregnancies have been screened to compare the results of a formal oral glucose tolerance test with timed venous plasma glucose measurements taken throughout the day. At the suggested thresholds of 6.1 mmol/l (within 2 h of eating) and 5.6 mmol/l (at greater than 2 h of food), random blood glucose testing is specific and excludes most normal women. However, the sensitivity of the test is low, and the majority of those with impaired glucose tolerance would be missed. Random blood glucose measurement is not a sufficiently sensitive method for detecting gestational diabetes as presently defined.