The driving force for organo- or lithotrophic growth as well as for each step in the metabolic network is the Gibbs reaction energy. For each enzymic step it must be neg. Thermodn. contributes therefore to the in-silico description of living systems. It may be used for assessing the feasibility of a given pathway because it provides a further constraint for those pathways which are feasible from the point of view of mass balance calcns. (metabolic flux anal.) and the genetic potential of an organism. However, when this constraint was applied to lactic acid fermn. according to a method proposed by Mavrovouniotis (1993a, ISMB 93:273-283) it turned out that an unrealistically wide metabolite concn. range had to be assumed to make this well-known glycolytic pathway thermodynamically feasible. During a search for the reasons of this surprising result the insufficient consideration of the activity coeffs. was identified as main cause. However, it is shown in the present contribution that the influence of the activity co-efficients on Gibbs reaction energy can be easily taken into account based on the intracellular ionic strength. The uncertainty of the tabulated equil. consts. and of the apparent std. Gibbs energies derived from them was found to be the second most important reason for the erroneous result of the feasibility anal. Deviations of intracellular pH from the std. value and bad estns. of currency metabolites, e.g., NAD+ and NADH, were found to be of lesser importance but not negligible. The pH dependency of Gibbs reaction enthalpy was proved to be easily taken into account. Therefore, the application of thermodn. for a better in-silico prediction of the behavior of living cell factories calls predominantly for better equil. data detd. under well defined conditions and also for a more detailed knowledge about the intracellular ionic strength and pH value. [on SciFinder (R)]