In many arm or leg movements the hand or foot has to exert an external force on the environment. Based on an inverse dynamical analysis of cycling, it is shown that the distribution of net moments in the joints needed to control the direction of the external force is often opposite to the direction of joint displacements associated with this task. Kinetic and kinematic data were obtained from five experienced cyclists during ergometer cycling by means of film analysis and pedal force measurement. An inverse dynamic analysis, based on a linked segments model, yielded net joint moments, joint powers and muscle shortening velocities of eight leg muscles. Activation patterns of the muscles were obtained by means of surface electromyography. The results show that the transfer of rotations in hip, knee and ankle joints into the translation of the pedal is constrained by conflicting requirements. This occurs between the joint moments necessary to contribute to joint power and the moments necessary to establish a direction of the force on the pedal which allows this force to do work on the pedal. Co-activation of mono-articular agonists and their bi-articular antagonists appear to provide a unique solution for these conflicting requirements: bi-articular muscles appear to be able to control the desired direction of the external force on the pedal by adjusting the relative distribution of net moments over the joints while mono-articular muscles appear to be primarily activated when they are in the position to shorten and thus to contribute to positive work. Examples are given to illustrate the universal nature of this constrained control of force (external) and position (joint). Based on this study and published data it is suggested that different processes may underlie the organization of the control of mono- and bi-articular muscles.