Six longitudinal data sets were used to examine the consequences of retirement controlling for preretirement characteristics. Results show (a) about one-half to three-fourths of income differences between retired and working men was caused by retirement; (b) little, if any, of the health differences are caused by retirement; (c) there are few effects of retirement on social activity; and (d) there are few effects on attitudes such as life satisfaction and happiness. Early retirement, however, has stronger effects than retirement at normal ages. The results show that retirement has different effects depending on type of outcome and timing of retirement.