Load in the human body can be quantified as force, stress, or strain, depending on the anatomical structure and the measuring technique. Direct measurements of these variables are invasive and only possible in animals or in small-scale in vivo studies in humans. Miniaturization of transducers and electronics may open new possibilities for direct measurements of load in the human body. Studies with a large number of human subjects, and routine analysis of patients, are done using noninvasive techniques: EMG analysis for muscle forces, kinematic analysis for ligament forces, and inverse dynamics for resultant joint loads. Inverse dynamics is the most general method and is applicable to all joints in the human body. Important limitations of inverse dynamics are due to the "distribution problem": the separation of resultant loads into the individual forces in muscles and other structures. Dynamic optimization is the most promising solution method for this problem. Inverse dynamics also relies heavily on the assumption that body segments are rigid. The errors caused by this simplification are most severe in impact and vibration studies. Computer simulation is a well-established method for load analysis in mechanical engineering but is relatively rare in biomechanics. Replacing the human test subject by a mathematical model has many advantages, mainly for reproducibility and understanding of the results. Models for mechanical properties and control of muscles are an important and difficult part of computer simulation. For this reason, computer simulation has only been applied for load analysis in impact simulations, where the muscles can be regarded as passive, or for certain special problems where similarly simple muscle models can be used. In the future, we may see more applications of computer simulation for analysis of more complex activities, such as gait and sports.