A higher inoculum size of beta-lactamase-positive Haemophilus influenzae is reported to increase minimum inhibitory concentrations (MICs) for beta-lactams. However, the effect of inoculum size of beta-lactamase-negative, ampicillin-resistant H. influenzae (BLNAR) on MICs for carbapenems has not been investigated. This study evaluated the effect of inoculum size on MICs for carbapenems and other beta-lactams in nine clinical isolates of BLNAR. The MICs were determined by both the standard method described by the Clinical and Laboratory Standards Institute (final inoculum size of 5 x 10(5) colony-forming units [CFU]/ml) and a modified method (final inoculum size of 5 x 10(6) CFU/ml) using viable cell counts. The findings showed that the higher inoculum size increased MICs for imipenem, meropenem, panipenem, biapenem, ampicillin, ceftazidime, and ceftriaxone. The inoculum effect (4 log(2) dilution or a greater increase in the MIC) with imipenem, meropenem, panipenem, and biapenem was found in three, five, two, and two isolates, respectively. The magnitude of the inoculum effect for panipenem significantly increased with the levels of MICs, but correlation between them for the others was not statistically significant. The mutations of penicillin-binding protein genes had little relevance to the reduced susceptibility to carbapenems or to the magnitude of the inoculum effect. These results suggest that MIC determination using turbidity can produce interpretive errors in the antimicrobial susceptibility testing of BLNAR for carbapenems because of their inoculum effect. Thus, accurate adjustment of inoculum size, such as viable cell count, is helpful for confirming the true MICs when the isolates are interpreted as "resistant" by turbidity-based MIC determination.