There is increasing interest among policy makers in measuring well-being in ways that go beyond purely economic indicators, also with special focus on older individuals who constitute an increasing fraction of the population.  However there is little consensus on which other indicators should be included.  An alternative approach is to use individuals’ own assessments and relate these to a rich set of covariates to find what factors influence individuals’ own perceptions.  This is the approach adopted in this paper, using data from the Health and Retirement Study (HRS).  Retired respondents are asked how satisfying their retirement has turned out to be, how retirement years compare to pre-retirement years and whether they are worried about not having enough income to get by in retirement.  I relate these self-assessed measures to a rich set of covariates to investigate which aspects weigh in individuals’ perceptions.  I use the longitudinal nature of the HRS to study the pathways that lead up to the observed retirement outcomes, and to examine the persistence of the outcomes over time.  Bad health, changes towards worse health, social isolation and increase in social isolation lead most significantly to lower satisfaction in retirement and a greater sense of financial insecurity in retirement. A short financial planning horizon and past shocks, like unexpected large expenses or divorce, also have a noticeable negative impact.