The life-cycle hypothesis implies that consumption would not decline at retirement. However, several studies found relevant declines in food consumption after retirement for the United States. Others concluded that this contradiction of the life-cycle hypothesis is solved by allowing for broader measures of consumption than food. Using repeated crosssection data for Germany, this paper analyzes the retirement consumption puzzle for the German case. For our broadest consumption measure, which includes the flow of durables’ consumption, we find, on average, no significant consumption decline at retirement. This also holds if the potential endogeneity of indidual retirement is controlled for in instrumental variable regressions. We also find heterogeneity in retirement effects among birth cohorts, the level of household wealth, and the level of consumption, but these effects do not support the hypothesis that retirement is associated with a strong reduction of consumption among poorer households.