This paper discusses specification and interpretation of human capital models of women's earnings when data on actual work experience are available. It uses the segmented earnings function framework developed by Jacob Mincer and Solomon Polachek and considers the effects of data errors, issues involving data interpretation, consequences of model mis-specification, and the simultaneity problem. The male-female wage gap is reexamined in light of these considerations. It is concluded that appropriate analysis of the corrected data does not support originally published conclusions concernig the shape of the on-the-job investment profile, the importance of "depreciation" of human capital, and the proportion of wage gap that can be explained by differences in work experience between men and women. (Author/JT)