In order to develop indices of social activities for the elderly, two surveys with 2 year intervals were conducted on the same 5,201 elderly subjects in four areas in Japan using a self-administered questionnaire. Social activities were defined as "activities which required contact with society" and were measured by 4 major facets of social activities, which were based on 21 questions relating to job activity, socially-plated activities, learning activities, and personal activities. The results were as follows; 1. The Wilcoxon scores in indices for 4 facets were given in sex and age groups. 2. Means of scores of indices increased with the degree of social activities from a subjective judgment. 3. Rank correlation coefficients between indices in two surveys were 0.60-0.71 for the persons whose answers were "no" to the question "did degrees of your activities change over the two years?" 4. Differences between indices in two surveys were higher in the persons with answers of "increase" to the above question than those with answers of "no", and were lower in those with answers of "decrease". These findings suggest that indices are available for assessing social activities as indicated by the reproducibility, validity and responsiveness found in this study.