In the absence of typical exclusion restrictions, covariance restrictions are used to obtain estimates of the effects of children on household behaviour. Using data from the PSID on two age samples, children are found to have a significant impact on many household decisions. However, while in the young sample exogenous fertility cannot be rejected, in the older sample this is not the case. Finally, if the average household had one less child, the male-female wage differential would decrease by 9.5 %.