We model the action of muscle-tendon system(s) about a given joint as a serial actuator and spring. By this technique, the experimental joint moment is imposed while the combined angular deflection of the actuator and spring are constrained to match the experimental joint angle throughout the stance duration. The same technique is applied to the radial leg (i.e., shoulder/hip-to-foot). The spring constant that minimizes total actuator work is considered optimal, and this minimum work is expressed as a fraction of total joint/radial leg work, yielding an actuation ratio (AR; 1 = pure actuation and 0 = pure compliance). To address work modulation, we determined the specific net work (SNW), the absolute value of net divided by total work. This ratio is unity when only positive or negative work is done and zero when equal energy is absorbed and returned. Our proximodistal predictions of joint function are supported during level and 15 degrees grade running. The greatest AR and SNW are found in the proximal leg joints (elbow and knee). The ankle joint is the principal spring of the hindleg and shows no significant change in SNW with grade, reflecting the true compliance of the common calcaneal tendon. The principal foreleg spring is the metacarpophalangeal joint. The observed pattern of proximal actuation and distal compliance, as well as the substantial SNW at proximal joints, minimal SNW at intermediate joints, and variable energy absorption at distal joints, may emerge as general principles in quadruped limb mechanics and help to inform the leg designs of highly capable running robots.