Although the NCCLS has approved the agar dilution method as the test of choice for antimicrobial susceptibility testing of Helicobacter pylori, a critical evaluation of this method in clinical trials to detect antibiotic resistance has not been performed. This study compares the Etest and agar dilution methods for detection of metronidazole and clarithromycin resistance in clinical isolates of H. pylori. MIC data were gathered from US-based clinical trials. The Etest was performed on Mueller-Hinton sheep blood agar plates following incubation for 4 days under 12% CO(2). The agar dilution test was performed according to the recently approved NCCLS methodology using aged sheep blood in a Mueller-Hinton agar base. Metronidazole resistance as determined by Etest was significantly higher than that determined by agar dilution (39%; 690/1768 vs. 25. 1%; 367/1465)(P<0.01). Clarithromycin resistance as determined by Etest was higher than that determined by agar dilution, but was not significantly different (12.5%; 209/1671 vs. 10.6%; 150/1414)(P>0.5). Inter-patient metronidazole resistance showed that the MIC values for identical isolates tested by both methods were equivalent in 58% (109/188). Of the 42% with a >2log(2) difference in MIC values, 17. 6% had a change in susceptibility pattern. For clarithromycin, 71.4% (237/332) of the MIC values for identical isolates tested by both methods had equivalent MIC values. Of the MIC values with a >2log(2) difference in MIC values, only 3% showed a change in susceptibility pattern. Intra-patient variability, i.e. paired isolates from the same patient, was assessed only for metronidazole. Of the 1393 paired isolates tested by Etest, 38.8% were shown to be resistant. Almost 69% of the Etest MIC determinations were deemed equivalent and 16.7% had a change in susceptibility pattern. Of the 639 paired isolates tested by agar dilution, 23.9% were resistant to metronidazole. Almost 72% of the agar dilution MIC values were equivalent and 11.3% of the determinations had a change in susceptibility pattern. Clarithromycin resistance rates are similar, when determined by either test method. The Etest yields a significantly higher prevalence of metronidazole resistance among H. pylori compared with the agar dilution method and both methods yield discordant results, when isolates from different parts of the same stomach are compared. Neither method is reliable in determining metronidazole resistance in H. pylori.