A recent investigation indicates that rapid antimicrobial susceptibility tests (AST) can affect patient therapy leading to reductions in health-care costs for some patient populations. However, there is little information relative to the often performed direct inoculation of positive blood culture bottles into rapid AST systems. AST results of direct inoculated Vitek (bioMerieux Vitek, Hazelwood, MO, USA) GNS cards were compared to those inoculated per package insert recommendations and a reference broth microdilution test using 50 consecutive Enterobacteriaceae bloodstream infection isolates. Escherichia coli (44% of isolates), Klebsiella ssp. (30%), and six other members of this family were tested against 15 antimicrobial agents. The direct inoculation method produced only two false-susceptible (0.3%), seven false-resistant (0.9%; six different drugs), and 48 minor errors (6.4%). The GNS cards inoculated in the usual, recommended manner had no very major error, and 7.5% combined major and minor errors. If the results of the urinary infection-specific drugs (nitrofurantoin, trimethoprim/sulfamethoxazole; not appropriate for bacteremia therapy) and ampicillin/sulbactam were deleted, both Vitek inoculation methods yielded results well within acceptable limits (< or = 4.5% overall error). These results indicate that the direct inoculation method of Vitek GNS cards from Enterobacteriaceae bloodstream infections (detected by Bactec 9240, Becton-Dickinson, Cockeysville, MD, USA) performed as well as the NCCLS broth microdilution test. Thus, a procedural modification of this type could further accelerate rapid access to accurate AST data.