In many microorganisms, flux limitations in oxidative metab. lead to the formation of overflow metabolites even under fully aerobic conditions. This can be avoided if the specific growth rate is controlled at a low enough value. This is usually accomplished by controlling the substrate feeding profile in a fed-batch process. The present work proposes a control concept which is based on the online detection of metabolic state by online calcn. of mass and elemental balances. The advantages of this method are: the check of measurement consistency based on all of the available measurements, the min. requirement of a priori knowledge of metab., and the exclusive use of simple and established online techniques which do not require direct measurement of the metabolite in question. The control concept has been linked to a simple adaptive controller and applied to fed-batch cultures of Saccharomyces cerevisiae and Escherichia coli, organisms which express different overflow metabolites, ethanol and acetic acid, resp. Oxidative and oxidoreductive states of S. cerevisiae and E. coli cultures were detected with high precision. As demonstrated by the formation of acetic acid in E. coli cultures, metabolic states could be correctly distinguished for systems for which traditional methods, such as RQ, are insensitive. Hence, it could be shown that the control concept allowed avoidance of overflow metabolite formation and operation at max. oxidative biomass productivity and oxidative conversion of substrate into biomass. Based on mass and elemental balances, the proposed method addnl. provides a richness of addnl. information, such as yield coeffs. and estn. of concns. and specific conversion rates. These data certainly help the operator to addnl. evaluate the state of the process online. [on SciFinder (R)]