We use a unique dataset, containing individual survey data from the English Longitudinal Study of Ageing linked to administrative data on earnings histories from administrative records, to construct measures of lifetime earnings and examine how these relate to financial resources in retirement. Retirement income and wealth at retirement is, as expected, positively correlated with lifetime earnings but there is also substantial dispersion in retirement income and retirement wealth among people with similar lifetime earnings. For example, we find that those with greater numerical ability and higher education tend to have greater retirement resources even after controlling for differences in lifetime earnings. The retirement resources of single women are far less well explained by their own lifetime earnings than those of couples or single men. We hypothesise that, as the vast majority of single women in the age group considered had previously been married and are now widowed or divorced, this reflects the fact that we do not observe the lifetime earnings of their former spouses.