Effects of temperature and temperature changes on circadian clocks in cyanobacteria, unicellular algae, and plants, as well as fungi, arthropods, and vertebrates are reviewed. Periodic temperature with periods around 24 h even in the low range of 1-2 degrees C (strong Zeitgeber effect) can entrain all ectothermic (poikilothermic) organisms. This is also reflected by the phase shifts-recorded by phase response curves (PRCs)-that are elicited by step- or pulsewise changes in the temperature. The amount of phase shift (weak or strong type of PRC) depends on the amplitude of the temperature change and on its duration when applied as a pulse. Form and position of the PRC to temperature pulses are similar to those of the PRC to light pulses. A combined high/low temperature and light/dark cycle leads to a stabile phase and maximal amplitude of the circadian rhythm-when applied in phase (i.e., warm/light and cold/dark). When the two Zeitgeber cycles are phase-shifted against each other the phase of the circadian rhythm is determined by either Zeitgeber or by both, depending on the relative strength (amplitude) of both Zeitgeber signals and the sensitivity of the species/individual toward them. A phase jump of the circadian rhythm has been observed in several organisms at a certain phase relationship of the two Zeitgeber cycles. Ectothermic organisms show inter- and intraspecies plus seasonal variations in the temperature limits for the expression of the clock, either of the basic molecular mechanism, and/or the dependent variables. A step-down from higher temperatures or a step-up from lower temperatures to moderate temperatures often results in initiation of oscillations from phase positions that are about 180 degrees different. This may be explained by holding the clock at different phase positions (maximum or minimum of a clock component) or by significantly different levels of clock components at the higher or lower temperatures. Different permissive temperatures result in different circadian amplitudes, that usually show a species-specific optimum. In endothermic (homeothermic) organisms periodic temperature changes of about 24 h often cause entrainment, although with considerable individual differences, only if they are of rather high amplitudes (weak Zeitgeber effects). The same applies to the phase-shifting effects of temperature pulses. Isolated bird pineals and rat suprachiasmatic nuclei tissues on the other hand, respond to medium high temperature pulses and reveal PRCs similar to that of light signals. Therefore, one may speculate that the self-selected circadian rhythm of body temperature in reptiles or the endogenously controlled body temperature in homeotherms (some of which show temperature differences of more than 2 degrees C) may, in itself, serve as an internal entraining system. The so-called heterothermic mammals (undergoing low body temperature states in a daily or seasonal pattern) may be more sensitive to temperature changes. Effects of temperature elevation on the molecular clock mechanisms have been shown in Neurospora (induction of the frequency (FRQ) protein) and in Drosophila (degradation of the period (PER) and timeless (TIM) protein) and can explain observed phase shifts of rhythms in conidiation and locomotor activity, respectively. Temperature changes probably act directly on all processes of the clock mechanism some being more sensitive than the others. Temperature changes affect membrane properties, ion homeostasis, calcium influx, and other signal cascades (cAMP, cGMP, and the protein kinases A and C) (indirect effects) and may thus influence, in particular, protein phosphorylation processes of the clock mechanism. The temperature effects resemble to some degree those induced by light or by light-transducing neurons and their transmitters. In ectothermic vertebrates temperature changes significantly affect the melatonin rhythm, which in turn exerts entraining (phase shifting) functions.