Multiparameter flow cytometric techniques developed in our laboratories have been used for the “at-line” study of fed-batch bacterial fermentations. These fermentations were done at two scales, production (20 m3) and bench (5 × 10−3 m3). In addition, at the bench scale, experiments were undertaken where the difficulty of achieving good mixing (broth homogeneity), similar to that found at the production scale, was simulated by using a two-compartment model. Flow cytometric analysis of cells in broth samples, based on a dual-staining protocol, has revealed, for the first time, that a progressive change in cell physiological state generally occurs throughout the course of such fermentations. The technique has demonstrated that a changing microenvironment with respect to substrate concentration (glucose and dissolved oxygen tension [DOT]) has a profound effect on cell physiology and hence on viable biomass yield. The relatively poorly mixed conditions in the large-scale fermentor were found to lead to a low biomass yield, but, surprisingly, were associated with a high cell viability (with respect to cytoplasmic membrane permeability) throughout the fermentation. The small-scale fermentation that most clearly mimicked the large-scale heterogeneity (i.e., a region of high glucose concentration and low DOT analogous to a feed zone) gave similar results. On the other hand, the small-scale well-mixed fermentation gave the highest biomass yield, but again, surprisingly, the lowest cell viability. The scaled-down simulations with high DOT throughout and locally low or high glucose gave biomass and viabilities between. Reasons for these results are examined in terms of environmental stress associated with an ever-increasing glucose limitation in the well-mixed case. On the other hand, at the large scale, and to differing degrees in scale-down simulations, cells periodically encounter regions of relatively higher glucose concentration. © 2000 John Wiley & Sons, Inc. Biotechnol Bioeng 70: 381–390, 2000.