Cross-province growth regressions for China are estimated for the reform period. Two research questions are asked. Can the regressions help us to understand why China as a whole has grown so fast? What types of investment matter for China's growth? We address the problem of model uncertainty by adopting two approaches to model selection to consider a wide range of candidate predictors of growth. Starting from the baseline equation, the growth impact of physical and human capital is examined using panel data techniques. Both forms of capital promote economic growth. ‘Investment in innovation’ and private investment are found to be particularly important. Secondary school enrolment contributes to growth, and higher education enrolment even more so.