Broth microdilution testing of 702 community-acquired isolates of Haemophilus influenzae from across Canada was performed with both Mueller-Hinton broth supplemented with 3% lysed horse blood broth (LHB) (BBL Microbiology Systems, Cockeysville, Md.) and haemophilus test medium (HTM). The prevalence of beta-lactamase production was found to be 26% with no regional variation. MICs determined with LHB tended to be higher than those with HTM, but interpretive errors due to these differences were observed only rarely with trimethoprim-sulfamethoxazole (n = 5), cefaclor (n = 8), and cefamandole (n = 3). The interobserver variability in MIC determinations was found to be greater when LHB was used than when HTM was used. There was no difference in intraobserver variability between the two medium formulations. beta-Lactamase-positive isolates developed false resistance to amoxicillin-clavulanate 2 weeks after microdilution panels of both types of medium were stored at -20 degrees C but not when panels were stored at -70 degrees C. In conclusion, this study supports the use of HTM rather than LHB for sensitivity testing of H. influenzae because of its lower rate of interobserver variability and its ability to support the growth of these organisms, which is comparable to that of LHB.