One hundred and fifty women who brought a home-voided urine sample to a general practitioner were asked at the same consultation to provide another sample under controlled conditions. Seventy-three women had lower urinary tract symptoms and 77 were asymptomatic. The two samples were analysed for pyuria (greater than or equal to 5 leucocytes/HPF) and bacteriuria (greater than or equal to 10(4) v. greater than or equal to 10(5) cfu/ml). The agreement when diagnosing pyuria and bacteriuria was estimated by Kappa (K). The home-voided sample was taken with adequate sampling technique by only 12% of women. Moreover, its bladder incubation time was twice that of the surgery sample (5.4 h v. 2.4 h). There was, however, no difference in the distribution of pyuria or bacteriuria between the paired samples. Agreement, which was high for the diagnosis of pyuria (K = 0.80), was significantly lower for the diagnosis of bacteriuria (K = 0.52, P less than 0.01). A home-voided sample seems sufficient in the diagnosis of the dysuria-pyuria syndrome in women.