A 4-h Micro-ID technique for direct identification of oxidase-negative gram-negative rods from positive blood cultures was compared to subculture and species identification of single colonies by API 20E and Micro-ID, using standardized inocula. A total of 127 patients (220 positive cultures) were studied. Isolates included 96 Escherichia coli, 46 Klebsiella pneumoniae, 7 Klebsiella oxytoca, 8 Enterobacter aerogenes, 17 Enterobacter cloacae, 19 Serratia marcescens, 2 Serratia liquefaciens, 8 Proteus mirabilis, 1 Salmonella species, 1 Morganella morganii, 6 Haemophilus influenzae, 2 Haemophilus parainfluenzae, 3 Bacteroides fragilis, 3 Acinetobacter calcoaceticus biotype anitratus, and 1 Pseudomonas maltophilia. In 90% of the cultures, identification by Micro-ID was identical to that obtained after subculture; if the 15 non-enterobacterial isolates were excluded, the corresponding figure was 96.6%. Enterobacteria identified incorrectly by direct Micro-ID were three S. marcescens (two identified as S. liquefaciens, one as Hafnia alvei), two S. liquefaciens (both identified as E. cloacae), and two K. pneumoniae (one identified as Klebsiella ozaenae, the other as Serratia rubidaea). None of the 15 non-enterobacterial cultures were correctly identified by Micro-ID (non-identifiable, or classified as Providencia/Yersinia/Klebsiella species). Although biochemical discrepancies between direct and final Micro-ID tests occurred in 41% of the enterobacterial cultures, this did not seriously interfere with identification. Direct species identification of Enterobacteriaceae from blood cultures by direct Micro-ID is accurate and easily performed and identified organisms within 4 h compared to at least 24 h by most other methods; the direct Micro-ID technique would be rendered even more valuable by the additional capability of identifying non-enterobacterial gram-negative isolates.