Precipitation is a primary climatic determinant of grassland productivity, with many global change experiments manipulating precipitation. Here we examine the impacts of precipitation addition and reduction treatment intensity and duration on grassland above- (ANPP) and below- (BNPP) ground net primary productivity in a large-scale meta-analysis. We tested, 1) the double asymmetry model of sensitivity, specifically whether the sensitivity of productivity decreases with treatment intensity under increased precipitation and increases with treatment intensity under decreased precipitation, 2) whether the sensitivity of productivity to precipitation change decreases with treatment length, and 3) how the sensitivity of productivity changes with climate conditions. ANPP showed higher sensitivity than BNPP under increased precipitation but similar sensitivity to BNPP under decreased precipitation. The sensitivity of ANPP and BNPP decreased with increasing treatment intensity (e.g., percentage change in precipitation, ΔPPT) and leveled off in the long-term. With increased precipitation, the sensitivity of productivity decreased with increasing treatment length (e.g., experimental duration) and leveled off in the long-term, whereas the sensitivity increased with increasing treatment length under reduced precipitation. Furthermore, the sensitivity of productivity to precipitation change decreased with increasing mean annual precipitation and temperature. Finally, our meta-analysis shows that above- and belowground net primary productivity have asymmetric responses to precipitation change. Together these results highlight the complex mechanisms underlying the impacts of precipitation change, particularly the intensity and duration of such changes, on grassland productivity.