Isokinetic exercise has become an increasingly popular modality in rehabilitative medicine during the past decade. The facility by which isokinetic dynamometers provide information about dynamic muscle contractions has, no doubt, been a major factor in this popularity. Isokinetic dynamometers are passive devices which resist applied forces and control the speed of exercise at a predetermined rate. Such dynamometers generally provide a record of applied force throughout a joint range of motion. Some of the postulated advantages of isokinetic exercise include safety, accommodating resistance and the facility for muscle force analysis. In spite of the advantages that isokinetic dynamometry provides, there are a number of considerations that are important in the interpretation of force recordings. While the term "isokinetics" generally denotes a type of muscular contraction which accompanies a constant rate of limb movement, periods of acceleration and deceleration exist in the context of isokinetic exercise. The acceleration and subsequent oscillatory and deceleration periods of "isokinetic" exercise limit the duration of the period of constant velocity in exercises with isokinetic dynamometry. Impact artifacts on torque records result from the compliance of the dynamometer system as it adjusts the accelerating limb to the present speed. The position at which peak torque occurs in a joint range varies with speed of motion. Therefore, analysis of maximal values at specific joint angles across speeds should be made in addition to the peak values generated throughout a joint range. The shape of the isokinetic force-velocity curve differs from the classic curve derived from prepared muscle specimens. As speed approaches zero, the isokinetic muscular force tends to rise much less steeply than that of the in vitro curve. Because measurements of absolute maximal force or velocity are subject to constraints in human studies, the direct comparison of in vivo to in vitro force-velocity curves is not justified. Neural inhibition of the force generated in intact muscle as tension rises has been postulated as a possible mechanism retarding the force curve at slow speeds. The reliability of specific types of isokinetic dynamometers appears to be quite high when test-retest analyses are performed with inert weights. Whether submaximal or maximal warm-ups are essential to ensure stable measures is still questionable at this time. It seems prudent to recommend submaximal warm-ups prior to maximal testing in order to reduce the possibility of muscle strain.(ABSTRACT TRUNCATED AT 400 WORDS)