The current research investigated the variability of school effects on intelligence development in considering two economically and socially distinct groups of children. The data came from a nationally representative sample of primary school children from urban and rural areas of China. Two standardised reasoning tests were used to assess fluid intelligence. The between-grade regression discontinuity approach was applied to disentangle the schooling effect from the age effect. Results showed that the combined effects of schooling and age were 5.09 intelligence points per year, in which the schooling effect (4.37 points) was larger than the age effect (0.72 points). The patterns of schooling and age effects were different in urban and rural subsamples. For the urban subsample, the schooling effect was smaller than the age effect for children from grade 1 to 3, but the schooling effect became larger than the age effect from grade 4 to 6. For the rural subsample, however, the schooling effect was always larger than the age effect. The effect of one year of schooling for rural children (5.59 points) was larger than that for urban children (3.15 points). These results suggest that schooling effects are closely related to the context where children are growing up.