This paper investigates the sources of rising lifetime incomes in Ghana using both existing cross-section data and a new panel data set which allows us to measure earnings growth within jobs and to construct lifetime work histories for both wage employees and the selfemployed. In Ghana - as elsewhere in sub-Saharan Africa - the nonrural self-employed constitute the most rapidly growing part of the urban labour force. Cross-sectional estimates of Mincerian wage earning equations show a concave relationship between experience and earnings, as is found in virtually all datasets, and a strongly convex relationship between education and earnings. We establish that similar crosssection age and tenure patterns exist for both the self-employed and wage employees. We then investigate the sources of these age-earnings profiles using time spent working, rather than age, as our measure of experience and testing if the cross-section data reflect within-job earnings growth. We find evidence of substantial bias in the cross-section estimates. The implications of these biases for earnings growth of the self-employed relative to wage employees are discussed.