AIMS--To compare the sensitivity and specificity of two semiautomated systems against a conventional (MIC 2000) test system for the identification and antibiotic susceptibility of Gram negative bacteria. METHODS--Clinical isolates of Gram negative bacilli (188 urinary and 229 non-urinary strains) were identified and tested for antibiotic susceptibility in the Cobas Micro and MIC 2000 systems. Of these, 359 strains were then tested in the Vitek and MIC 2000 systems. Two hundred and forty three strains were tested in all three systems immediately after isolation. Forty three were also tested only in the Vitek and MIC 2000 systems immediately after isolation. The remaining 174 strains were tested after storage at -20 degrees C for several months. RESULTS--The Cobas Micro and MIC 2000 systems agreed on the identification of 310 of the 417 (74.3%) strains; the Vitek and MIC 2000 systems agreed on 338 of the 359 (94.2%) strains. The Cobas Micro system correctly identified 86.8% of strains tested after storage and 65.4% of those immediately after isolation. Organism-antibiotic combinations (non-urinary isolates) were tested in the Cobas Micro and MIC 2000 systems (n = 2335), in the Vitek and MIC 2000 systems (n = 999). Essential correlation (complete agreement plus minor errors) was observed in 98% (with 90% complete agreement) in the former and in 97% (with 86% complete agreement) in the latter. For the urinary isolates, 1949 organism-antibiotic combinations were analysed in the Cobas Micro and MIC 2000 systems where complete agreement was observed in 92% (with 3% very major discrepancies), for 1382 urinary organism-antibiotic combinations tested in the Vitek and MIC 2000 systems, the figures were 95% and 2%, respectively. CONCLUSIONS--The Vitek system is highly accurate in the identification and antibiotic susceptibility testing of Gram negative bacteria. The Cobas Micro system has many shortcomings in its identification of Gram negative rods, especially freshly isolated strains, but it is comparable with the Vitek system in antibiotic susceptibility testing.