This paper describes a controlled study of the effects of an unsupervised walking-programme in 108 volunteers from among 580 middle-aged employees in a Nottingham light industrial company. Objective measurements of physical condition and of customary activity were made on four occasions at 12-week intervals. The subjects were randomly allocated to three groups and encouraged to follow a walking-programme in the first, second or third periods of the study: no activity was prescribed in the other periods. Compliance with the programme was estimated from log-cards and from a week's pedometer record. Although there were no significant differences between the changes seen in subjects following the walking-programme and those in control subjects, there were significant small improvements in physical condition and modest increases in customary activity in a) those subjects actually completing the programme, and b) the subjects measured before and after their programme, pooled without regard to the period of measurement. At least some of the increased activity is maintained for 12 weeks after the programme. The marginal changes suggest that the intensity of the training-programme as experienced by the subjects was close to the threshold for maintenance of physical condition in this age-group of workers.