Global climate change is expected to result in a greater frequency of extreme weather, which can cause lag effects on aboveground net primary production (ANPP). However, our understanding of lag effects is limited. To explore lag effects following extreme weather, we applied four treatments (control, doubled precipitation, 4 °C warming, and warming plus doubled precipitation) for 1 year in a randomized block design and monitored changes in ecosystem processes for 3 years in an old-field tallgrass prairie in central Oklahoma. Biomass was estimated twice in the pretreatment year, and three times during the treatment and posttreatment years. Total plant biomass was increased by warming in spring of the treatment year and by doubled precipitation in summer. However, double precipitation suppressed fall production. During the following spring, biomass production was significantly suppressed in the formerly warmed plots 2 months after treatments ceased. Nine months after the end of treatments, fall production remained suppressed in double precipitation and warming plus double precipitation treatments. Also, the formerly warmed plots still had a significantly greater proportion of C4 plants, while the warmed plus double precipitation plots retained a high proportion of C3 plants. The lag effects of warming on biomass did not match the temporal patterns of soil nitrogen availability determined by plant root simulator probes, but coincided with warming-induced decreases in available soil moisture in the deepest layers of soil which recovered to the pretreatment pattern approximately 10 months after the treatments ceased. Analyzing the data with an ecosystem model showed that the lagged temporal patterns of effects of warming and precipitation on biomass can be fully explained by warming-induced differences in soil moisture. Thus, both the experimental results and modeling analysis indicate that water availability regulates lag effects of warming on biomass production.