The screening of women of childbearing age for haematuria, leukocyturia and proteinuria to detect urinary schistosomiasis can be confounded by several factors such as menstruation, pregnancy and genitourinary infections. We therefore undertook a study in an area endemic for Schistosoma haematobium in the United Republic of Tanzania to carry out the following: assess the sensitivity, specificity and predictive values--in women of childbearing age--of indirect indicators of urinary schistosomiasis, as measured by urine reagent strip readings; assess the predictive values of self-reported symptoms; and finally to estimate the morbidity attributable to S. haematobium. A total of 303 women (128 and 175, respectively, living in high- and low-risk sites) participated in the study. Haematuria was more frequent among women excreting S. haematobium eggs than among those who did not (65% versus 32%). The predictive potential of all indirect disease markers was poor in the highly endemic site, while in the sites with low endemicity the negative predictive values were high. Among infected women, 54% of haematuria could be attributed to S. haematobium, but for patients with more than 10 eggs/10 ml the attributable fraction rose to 70%. Symptoms of "bloody urine" and "pain while urinating" were recalled significantly more often by women living in the highly endemic site. On a population level, one-third of the self-reported cases with bloody urine could be attributed to urinary schistosomiasis. Screening of women of childbearing age for urinary schistosomiasis using urine reagent strips can be biased in two directions. The prevalence of S. haematobium will be overestimated if other causes of haematuria, such as reproductive tract infections, are highly endemic. On the other hand, women with light or very light infections will be missed and will not be treated. This is of concern because genital schistosomiasis, a possible risk factor for the transmission of HIV, occurs among women even with light infections.