Individual time preference determines schooling enrolment. Moreover, smoking behavior in early ages has been shown to be highly related to time preference rates. Insofar as discount rates are uncorrelated to ability, predicting school enrolment by discount rates can get rid of the ability bias in an earnings regression. Accordingly, we use smoking at age 16 as an instrument for schooling. Doing this for Austrian cross-sectional data, we find no evidence of ability bias in a simple earnings regression; the results are more mixed if family background is also included.