We propose a unique fuzzy system design for adjusting the cycle ergometer work-load to each individual's physical work capacity. We observed the physical work capacity by measuring the heart rate and a muscular-fatigue-related index. A set of fuzzy membership functions was determined for three different phases during a trial of exercise with a progressively increasing workload. Since the available objective data was limited in describing changes in physical activities, the fuzzy rules for workload control were designed to reflect the different types and different time-scales of physical information. We examined 18 elderly people and periodically redesigned die fuzzy inference for eight selected persons over a seven-month period.