The aetiology of exercise-associated muscle cramps (EAMC), defined as 'painful, spasmodic, involuntary contractions of skeletal muscle during or immediately after physical exercise', has not been well investigated and is therefore not well understood. This review focuses on the physiological basis for skeletal muscle relaxation, a historical perspective and analysis of the commonly postulated causes of EAMC, and known facts about EAMC from recent clinical studies. Historically, the causes of EAMC have been proposed as (1) inherited abnormalities of substrate metabolism ('metabolic theory') (2) abnormalities of fluid balance ('dehydration theory'), (3) abnormalities of serum electrolyte concentrations ('electrolyte theory') and (4) extreme environmental conditions of heat or cold ('environmental theory'). Detailed analyses of the available scientific literature including data from recent studies do not support these hypothesis for the causes of EAMC. In a recent study, electromyographic (EMG) data obtained from runners during EAMC revealed that baseline activity is increased (between spasms of cramping) and that a reduction in the baseline EMG activity correlates well with clinical recovery. Furthermore, during acute EAMC the EMG activity is high, and passive stretching is effective in reducing EMG activity. This relieves the cramp probably by invoking the inverse stretch reflex. In two animal studies, abnormal reflex activity of the muscle spindle (increased activity) and the Golgi tendon organ (decreased activity) has been observed in fatigued muscle. We hypothesize that EAMC is caused by sustained abnormal spinal reflex activity which appears to be secondary to muscle fatigue. Local muscle fatigue is therefore responsible for increased muscle spindle afferent and decreased Golgi tendon organ afferent activity. Muscles which cross two joints can more easily be placed in shortened positions during exercise and would therefore decrease the Golgi tendon organ afferent activity. In addition, sustained abnormal reflex activity would explain increased baseline EMG activity between acute bouts of cramping. Finally, passive stretching invokes afferent activity from the Golgi tendon organ, thereby relieving the cramp and decreasing EMG activity.