I apply propensity score matching to estimate the average and individual effects of 65 democratizations in the 1965-2011 period on long-term growth. The matching successfully establishes balance between the control and treatment groups in terms of pre-transition democracy, economic development, wars and political instability, geographical characteristics, and time of transition. A democratization is estimated to raise long-term growth by more than 0.7 percentage points on average, but the individual effects of democratizations are highly heterogeneous. Counterfactual growth acceleration is nominal and even slightly negative, suggesting that prolonged autocracy harms growth. We, however, must be cautious in concluding that democracy benefits long-term growth.