Twelve laboratories in different parts of Britain each supplied approximately 80 consecutive urinary bacterial isolates from community patients. All strains were identified by a central laboratory, where sensitivity to a variety of orally administered antimicrobials was determined by microtitre broth dilution. 65.1% of isolates were Escherichia coli, 23.4% 'coliforms' other than E. coli, 4.6% Proteus and Morganella spp., 1.8% Pseudomonas spp., 2.4% enterococci, 0.7% group B streptococci, 1.5% coagulase-negative staphylococci and 0.5% Staphylococcus aureus. Using previously published breakpoint sensitivity values, 98.9% of all isolates were found to be sensitive to norfloxacin and to ciprofloxacin, 95.7% to co-amoxiclav, 86.8% to nitrofurantoin, 77.4% to cephalexin, 75.6% to trimethoprim, 75.0% to cephradine and 51.7% to amoxycillin. There were some differences in sensitivities between centres, particularly those of the cephalosporins. Using standard breakpoints, submitting laboratories were found to overestimate sensitivity to nitrofurantoin and to underestimate sensitivity to the quinolones and to co-amoxiclav; there was considerable overestimation of sensitivity to cephalosporins.