Using data from large scale, nationally representative sample surveys, we tested the hypothesis that prenatal exposure to famine increases schizophrenia risk at adulthood by studying the Great Leap Forward Famine in China (1959-1961). Our results show that, in the urban population, being conceived and born during the famine increased the risk of developing schizophrenia at early adulthood as compared to both the pre-famine and post-famine cohorts. In the rural population, however, the post-famine cohort had the highest risk of developing schizophrenia, and there was virtually no difference in schizophrenia risk between the pre-famine and the famine cohort. This finding contrasts sharply with previous studies on the Dutch Hunger Winter as well as with smaller scale local studies in China based on hospital records. We offer an explanation for the urban-rural difference in the schizophrenia-famine relationship based on population selection by differential excess mortality and provide supportive evidence through province- and cohort-level ecological analysis.