1. The relationship between [Na]o and the contracture tension, developed by isolated frog trabeculae, has been investigated in the presence of different levels of bathing K and Ca ions and after experimental manoeuvres likely to increase [Na]i. 2. Raising [K[o, [Ca]o or [Na]i all increase the strength of the contractures induced by lowering the bathing [Na] except when the [Na]o is close to zero, suggesting that the Ca-Na exchange depends on membrane potential or [K]o. 3. The experimental data have been compared to the predictions of three relatively simple models of Ca-Na exchange in the membrane, where [Ca]i depends on the [K]o either directly or by way of its effect on the membrane potential and tension varies according to a second order relationship with [Ca]i. 4. The only scheme to fit all the experimental data satisfactorily is one which assumes an exchange of one Ca ion for three Na ions across the membrane. This scheme requires that the contractile system has an apparent binding constant for Ca2+, similar to that found with skinned cardiac muscle cells or isolated cardiac myofibrils. In intact muscle, when the [Na]i is close to that measured experimentally, the model predicts that tension should vary with the [Ca]o2, 1/[Na]o4, and the [Ca]/[Na]o2 ratio.