1. Exchangeable Ca in guinea-pig auricles and ventricular trabeculae of sheep and calf hearts was labelled with (45)Ca and the loss of radioactivity into inactive rinsing solutions of different ion composition was measured for periods up to 6 hr. At no time did the decrease of radioactivity in the muscle follow a single exponential course, while the rate coefficient k (= fraction of (45)Ca lost per minute from muscle into rinsing solution) decreased slightly with time.2. On the basis of the temperature-sensitivity of Ca efflux from auricles the activation energy has been calculated to have a value of 5900 cal/mole, corresponding to a Q(10) of 1.35. 2,4-Dinitrophenol (5.5 x 10(-5) - 5.5 x 10(-4)M) had either no effect on Ca efflux or increased it slightly.3. Compared to control efflux in 1.8 mM [Ca](o) Ca efflux decreased to 70% in Ca-free solution, to 20% in Ca-free, Na-free solution and to 65% in Ca-containing, Na-free solution, NaCl being replaced by either sucrose or LiCl. Quantitatively, Ca efflux from auricles has been shown to depend to a large extent on the ratio [Ca(2+)](o)/[Na(+)](o) (2). The affinity for Na of the activation site for Ca efflux (carrier) is much less than for Ca.4. The efflux from muscles soaked for about 2 hr in Ca-free solution was not linearly related to Ca-concentration in the tissue but followed a square law.5. While Ca content in auricles increased in Ca-containing, Na-poor solution it decreased again when Tyrode solution was readmitted indicating a Na-sensitive Ca net transport in cardiac muscle.6. The results are interpreted in terms of a modified exchange diffusion mechanism (Ussing, 1947) which is responsible for Ca extrusion from mammalian cardiac muscle.