The inter-system reproducibility of the MIC results obtained from the MS-2 automated system of antibiotic sensitivity testing was assessed by a two-stage study. In the first stage, a total of 168 MICs were calculated by the automated system from 134 diverse bacterial strains, selected from routine antibiotic sensitivity testing for their "intermediate" sensitivity to one or more antibiotics. The antibiotic sensitivities were repeated using the same machine, in the same laboratory, on another suspension of the same bacteria. The results of antibiotic sensitivity reported as "intermediate" on the first test were compared to those of the second test; only 19 p. cent of the responses remained "intermediate" and the MICs were comparable between the two tests; 50 p. cent became sensitive and 31 p. cent became resistant (MIC not calculated). In the second stage, 4 strains with stable intermediate sensitivity to 8 antibiotics, as shown by 2 routine sensitivity tests, were tested 33, 48, 44, and 39 times respectively against these antibiotics using various suspensions of the same strain. 394 results were obtained: 92 intermediate results with a MIC (24.4 p. cent), 219 sensitive results (56.3 p. cent) and 83 resistant results (20.2 p. cent). The range of values for the MIC calculated by the automatic system was small, so the MICs are reproducible; but, for 5 of the 8 antibiotics, the responses were both "S" and "R". This suggests that the so-called "Intermediate" zone is in fact a zone of uncertainty where the majority of the MIC results are erroneous.