The objective of this study was to determine the accuracy of ward urinalysis and the sensitivity of dipstick testing in the assessment of proteinuria in hypertensive pregnant women. Subjects were 230 consecutive hypertensive pregnant women who were admitted to hospital over a 2-year period. Routine ward urinalyses for protein, obtained on a mid-stream sample before and after a 24-hour urine collection for quantitating proteinuria, were compared with the 24-hour urine protein excretion. As a control for dipstick accuracy, urinalysis was also performed on a mixed aliquot of each of the 24-hour samples by a single observer experienced in urinalysis. True proteinuria was considered as > 300 mg/day. The positive predictive value for urinalysis ranged from 38% (for the precollection test) to 60% (for tests on the aliquot). Negative predictive values were 86-88%. The false negative rates at 'nil' or 'trace' proteinuria ranged from 8-18%. The false positive rates at '3+' (3 g/L) or '4+' (> or = 20 g/L) ranged from 0-17%, at '2+' (1 g/L) from 18-50% and at '1+' (0.3 g/L) from 67-83%. Best results for urinalysis were obtained on the aliquot testing but even under these ideal circumstances there was a high false positive rate (67%) at '1+' (0.3 g/L) urinalysis level. These studies show that in routine clinical practice 'nil' or 'trace' proteinuria will miss significant proteinuria in approximately 1 out of 8 hypertensive pregnant women while '3+' (3 g/L) or '4+' ( > or = 20 g/L) will rarely be a false positive. At urinalysis of '1+' or '2+' a 24-hour urine collection is required to be certain about the presence or absence of proteinuria. Research studies should demand 24-hour urine protein quantitation and not rely solely upon urinalysis results.