Cross-sectional analysis is problematic when examining the determinants of migration as well as its impacts. Panel data may potentially solve the problem by tracking households over different time periods. Using panel data from household surveys in six provinces in rural China over 1986 to 1999, this paper examines the determinants of rural-to-urban migration and its impacts on rural households. We find that number of laborers, income, education level and village migrating network increase the likelihood of migration for households with no migration experience as well as households with experience. By estimating the dynamic difference on migration impacts, at the household level we find that grain output declines by less than 2 percent while net income increases by 16 percent upon migration.