Formulates a stochastic dynamic model of fertility to evaluate the assumptions that underlie the widely used econometric tests for parental sex preferences. Unlike previous work on dynamic models of fertility, several tractable and testable predictions are established. It is shown rigorously that conventional econometric tests using fertility data are valid tests for sex preferences; however, they cannot separate son preference from daughter preference. The only definite conclusion that one can draw from fertility data is whether there are sex preferences. These results call into question the validity of conventional econometric tests for son preference. -Author