Temperature is an environmental factor, which influences most of the chemical processes occurring in living and nonliving systems. Van't Hoff's rule states that reaction rates increase by a factor (the Q 10) of two or more when the temperature is increased by 10 °C [1]. Despite this strong influence of temperature on individual reactions, many organisms are able to keep some of their metabolic fluxes at an approximately constant level over an extended temperature range. Examples are the oxygen consumption rates of ectoterms living in costal zones [2] and of fish [3], the period lengths of all circadian [4] and some ultradian [5,6] rhythms, photosynthesis in cold-adapted plants [7,8], homeostasis during fever [9], or the regulation of heat shock proteins [10]. In 1957, Hastings and Sweeney suggested that in biological clocks such temperature compensation may occur as the result of opposing reactions within the metabolic network [11]. Later kinetic analysis of the problem [12] reached essentially the same conclusion, and predictions of the theory have been tested by experiments using Neurospora's circadian clock [13] and chemical oscillators [14,15]. In this study, we use metabolic and hierarchical control analysis [16-22] to show how certain steady-state fluxes in static reaction networks can be temperature compensated according to a similar principle, and how dynamic networks have an additional repertoire of mechanisms. This study is mostly theoretical, but we use the temperature adaptation of yeast cells and of photosynthesis as illustrations. These and other Temperature has a strong influence on most individual biochemical reactions. Despite this, many organisms have the remarkable ability to keep certain physiological fluxes approximately constant over an extended temperature range. In this study, we show how temperature compensation can be considered as a pathway phenomenon rather than the result of a single-enzyme property. Using metabolic control analysis, it is possible to identify reaction networks that exhibit temperature compensation. Because most activation enthalpies are positive, temperature compensation of a flux can occur when certain control coefficients are negative. This can be achieved in networks with branching reactions or if the first irreversible reaction is regulated by a feedback loop. Hierarchical control analysis shows that networks that are dynamic through regulated gene expression or signal trans-duction may offer additional possibilities to bring the apparent activation enthalpies close to zero and lead to temperature compensation. A calori-metric experiment with yeast provides evidence that such a dynamic temperature adaptation can actually occur.