Devices and methods have been developed for determining speed and oxygen cost of level walking. Speed was recorded and controlled using a speedometer cart. Oxygen uptake was determined with an argon-dilution method using a mixing box mounted on a backpack. The method was found valid, coefficient of variation (cv) less than 2.1%, and to give excellent reproducibility with regard to self-selected speed, cv less than 1.9%, predetermined speed, cv less than 1.3%, and to oxygen cost, cv less than 3.2%. Artificially arranged immobility of the knee or instability of the ankle decreased comfortable walking speed 23% and 4% respectively. Oxygen cost increased 23% and 10% respectively. Stabilizing splints allowing some flexion could if possible be advocated, particularly with elderly patients. Patients with coxarthrosis were studied before and after THA. One year after surgery the Harris hip score had increased from 35 to 85 points and maximal walking speed from 62 to 80 m min-1 Oxygen cost had decreased from 0.267 to 0.221 ml kg-1m-1. The onset of and the recovery from complications, as well as differences between patients with uni- and bilateral diseases, were reflected in change in oxygen cost but not in clinical scoring. Patients with moderate gonarthrosis were studied before and after unicompartmental knee prosthetic replacement. No major benefit of preoperative physical therapy, mainly aiming to improve thigh muscle strength, was observed three months after surgery. One year after surgery the patients had improved in clinical score rating and recovered an almost normal walking ability. Measurements of pain and self-selected walking speed were found to be sufficient for assessing effects of treatment in these patients. Patients with severe gonarthrosis had improved in clinical score rating one year after TKR. Oxygen cost of walking was unchanged. An acquired uneconomic walking pattern was considered to be the reason for unimproved walking efficiency. Patients with spastic paraparesis were treated with long-term stretch of the hip adductor muscles. Either the oxygen cost or the blood lactate level was decreased during walking, indicating that even during moderate exercise blood lactate must be taken into consideration when energy cost is measured in these patients. Measurements of walking speed and oxygen cost of level walking were found to be useful objective parametres for assessing walking and to be a valuable supplement to clinical assessment of effects of treatment in patients with walking disorders.