The energy cost for maintenance of gradients of methylthio-beta-d-galactoside in Escherichia coli was evaluated. Information was also obtained concerning the energy flow associated with gradient establishment under some circumstances. Energy flow was evaluated from transport-induced changes in the rate of heat evolution, oxygen consumption, and carbon dioxide production in metabolically active cells. Heats were measured with an isothermal calorimeter. Energy expenditure behavior was characterized by a transition that depended on the level of accumulation. The data for steady-state maintenance could be rationalized in terms of the Mitchell hypothesis, two models for influx and efflux, and a transition between them. At low levels of uptake, steady-state proton-methylthio-beta-d-galactoside (TMG) symport for influx and efflux occurred via a nonenergy-requiring exchange process. The only energy requirement was that necessary to pump back in any TMG exiting via a leakage pathway (model I). Above the transition, all influx occurred with proton symport, but all exit, leak and carrier mediated, occurred without proton symport (model II). The H(+)/TMG stoichiometric ratio computed for the region of model II applicability (carbon source present, high level of uptake) approached 1. This value agreed with that of other workers for downhill beta-galactoside flow, suggesting that the energy cost for both downhill and uphill flow was approximately the same. For low levels of uptake, initial establishment of the gradient was followed by a burst of metabolism that was much larger than that expected on the basis of the chemiosmotic hypothesis. In the absence of carbon source, the stimulation in respiration was sufficient to produce 13 times more protons than are apparently necessary to establish the gradient. The results indicate also that the nature of the biochemical process stimulated by TMG depends on its level of uptake. Insight into several aspects of the nature of these processes was provided through analysis of the heat, oxygen, and CO(2) data. The key factor controlling the transition in energy flow behavior is suggested to be rate of flux. The present data suggest that it occurs at a flux of approximately 120 nmol/min per mg of protein.