In this paper three models which describe the relation between earnings and education are estimated, using one single data set. One model is a simple dummy regression model in which the educational variables are exogenous. Both other models take account of the fact that individuals choose their optimal amount of education, and is therefore endogenous. In all three models I try to specify the equations in the same way as others have done with other data sets. It turns out that in each model important conclusions reached in earlier research, are not supported by the empirical findings reported here.