In this paper we use child level data from Pakistan to estimate the probability of child mortality. We find that overall girls have a higher probability of surviving and when we look at disaggregated data we find that relative to boys, girls have a significantly lower probability of dying in the age group 0-1 but have a significantly higher probability of dying in the age group 1-5. The results are robust to mother level unobserved heterogeneity. Education of the mother is seen to have a significant and negative effect on child mortality and there is a threshold level of education that the mother has to attain before education starts affecting child mortality. We also find that increased duration between the births significantly reduces child mortality. We argue that the higher mortality of girls in the age group 1-5 reflects discrimination against girls in the form of lower health and other resource inputs.