Agents that activate or inhibit the $Ca^{2+}$ release channel in cardiac sarcoplasmic reticulum (SR) were tested for their abilities to affect the activity of the SR $Ca^{2+}$-ATPase. Vesicles of junctional SR (heavy SR, HSR) from terminal cisternae were prepared from porcine cardiac muscle by density gradient centrifugation. The steady-state activity of $Ca^{2+}$-ATPases in intact HSR vesicles was/$347{pm}5;nmol/min{cdot}mg$ protein (${pm}$ SD). When the HSR vesicles were made leaky, the activity was increased to $415{pm}5;nmol/min{cdot}mg$ protein. This increase is probably due to the uncoupling of HSR vesicles. Caffeine (10 mM), an agonist of the SR $Ca^{2+}$ release channel, increased $Ca^{2+}$-ATPase activity in the intact HSR vesicle preparation to $394{pm}30;nmol/min{cdot}mg$ protein. However, caffeine had no significant effect in the leaky vesicle preparation and in the purified $Ca^{2+}$-ATPase preparation. The effect of caffeine on SR $Ca^{2+}$-ATPase was investigated at various concentrations of $Ca^{2+}$. Caffeine increased the pump activity over the whole range of $Ca^{2+}$ concentrations, from $1;{mu}M$ to $250;{mu}M$, in the intact HSR vesicles. When the SR $Ca^{2+}$-ATPase was inhibited by thapsigargin, no caffeine effect was observed. These results imply that the caffeine effect requires the intact vesicles and that the increase in $Ca^{2+}$-ATPase activity is not due to a direct interaction of caffeine with the enzyme. We propose that the activity of SR $Ca^{2+}$-ATPase is linked indirectly to the activity of the $Ca^{2+}$ release channel (ryanodine receptor) and may depend upon the amount of $Ca^{2+}$ released by the channels.