Taking' multidimensional life satisfaction as the basic premise of this study, a four-equation multiple regression model was constructed for its prediction. The set of regressors included some conventional biographical characteristics, a number of activity variables, and three indicators of retirement experience. A stepwise regression estimation method was adopted. Results indicated that the pattern of regressor influence varied greatly between equations, providing fairly specific evidence on a number of previously espoused hypotheses. In particular, the activity perspective on aging was found to be important in the prediction of over-all affect or mood tone, but of little relevance for the other dimensions of life satisfaction, and a modicum of support for the disengagement perspective was observed.