Measurements of protein/creatinine ratio in 'spot' urine samples were compared with measurements of 24 hour quantitative proteinuria and side room 'dipstick' testing in 104 samples from 90 patients presenting consecutively to a rheumatology unit. Linear regression analysis showed a highly significant correlation between the random urinary protein/creatinine ratio and total protein excretion in 24 hour urine samples (r = 0.92, p less than 0.001, y = 6.55x + 0.04). Although an approximation of 24 hour urinary protein excretion could have been made from the regression line: 24 hour urine protein = 6.55 x protein/creatinine ratio + 0.04 (g/l), there was a wide scatter of values, particularly in patients with greater than 1 g/24 h urinary protein excretion. Nevertheless, significant proteinuria (greater than 300 mg/24 h) could have been confirmed or excluded with a sensitivity and specificity of 97% by adopting random protein/creatinine values of less than 0.04 as 'normal'. Specificity and sensitivity could have been increased to 100%, however, by excluding patients with values lying between 0.01 and 0.10 as all the false negatives (n = 3) and false positives (n = 3) lay within this range. In comparison, dipstick testing, although 100% sensitive, had a poor specificity due to the high false positive rate (40/83 (48%] in patients with 1+ to 3+ readings. Assessment of random urinary protein/creatinine ratio may obviate the need for 24 hour urine collections in the initial assessment of suspected proteinuria. A wider application of this technique seems indicated in view of the obvious advantages in terms of cost, time, and patient convenience.