The practicalities of conducting exercise studies with elderly and very elderly people have not been well described. In order to help others plan and perform such studies we describe our experience of recruiting volunteers, applying selection criteria, measuring strength, power, cardiorespiratory responses, and potentially related functional abilities. Exclusion criteria are offered, for safety and to characterize subjects as free of disease which might alter their exercise performance. International agreement on these, or similar, criteria would be valuable. The budget must be adequate for prolonged recruitment before a study and for the liberal use of taxis during it. With healthy subjects in their seventies, the coefficients of variation (CV) for repeated measurements of strength and power were: handgrip 3%, isometric knee extension 6%, isometric elbow flexion 6%, and lower limb extensor power 9%. CV for isometric knee extension by healthy subjects in their eighties was 4%. Treadmill ergometry is more time-consuming than with younger subjects. During progressive treadmill tests, the heart rate interpolated to oxygen consumptions of 10 and 15 ml.kg−1.min−1 had CV=4% and 7%, respectively.