A total of 14,272 urine specimens were examined over one year to determine the validity of direct antimicrobial agent susceptibility testing against ampicillin, amoxicillin-clavulanic acid, cephalothin, gentamicin, norfloxacin, and trimethoprim. A comparison between direct and standardized disk diffusion tests was made for a total of 1,106 urine specimens containing > or = 10(5) organisms per ml in pure culture. There were 5,821 individual organism-antimicrobial agent challenges compared for the two testing methods, and there was complete agreement of susceptibility category in 5,492 comparisons (94.3%). Initially, discordant results were reduced from 5.7 to 2.1% when the intermediate category was considered susceptible. Intralaboratory variation was assessed by testing another 453 organisms by the standard National Committee for Clinical Laboratory Standards (NCCLS) method on two consecutive days; there was complete agreement in 96.1% of comparisons. When results of direct and standardized testing were simply classified as susceptible or resistant, there was 1.1% discordance. When simple same-day tests were used together with predictable patterns of susceptibility and resistance, 536 (48.5%) of 1,106 isolates could be identified satisfactorily to the genus or species level. For laboratory reporting purposes, the direct method is equivalent to the standard method when the urine being tested is infected with > or = 10(5) organisms of a single type per ml. The presence or absence of preexisting antimicrobial agents in urine did not appreciably influence the results. This procedure allows the earlier reporting of susceptibility results and facilitates less expensive identification of many organisms. Costs and benefits need to be determined in each institution.