We have developed a standard procedure to measure the susceptibility of Trichomonas vaginalis to metronidazole in vitro, by controlling the state of oxygenation of the test cultures, the size of the inocula in terms of infectivity, and providing multiple tests at each drug concentration. Estimates were made of the in-vitro sensitivity to metronidazole of T. vaginalis isolated from 11 randomly selected patients, all, except for three defaulters, known to have been treated successfully, and from six 'treatment failure' patients. Stocks isolated from 'successfully treated' patients, were all highly sensitive to metronidazole. Eighteen stocks isolated from the 'treatment failure' patients fell into two groups: (a) those clearly separable from sensitive stocks, with ED50 values (the concentration of drug at which growth is prevented in 50% of test wells) in N2 of up to 0.51 mg/l and 3.5 to 11 mg/l in N2/1%O2; (b) those giving intermediate ED50 values of up to 0.20 mg/l in N2 and 0.7 to 1.8 mg/l in N2/1%O2. The current procedure demonstrates clearly that repeated treatment failure in patients without other complication is associated with enhanced resistance to metronidazole of T. vaginalis, and it may discriminate between T. vaginalis infections that are likely to respond to higher dosage and those that are not.