Of the total land worldwide, 43 % is phosphorus (P)-constrained, which is significantly greater than the proportion of nitrogen-constrained land (18 %). However, there are some uncertainties regarding the responses of enzyme activities to P addition on a global scale. We collected data from 99 publications and performed a meta-analysis to assess the responses of 11 extracellular enzymes, including carbon, nitrogen, and phosphorus acquisition enzymes, and oxidases. P addition significantly increased the activities of β-1, 4-glucosidase, sucrase, cellobiohydrolase, and alkaline phosphatase, while lowering the activities of acid phosphatase and phosphodiesterase. The responses of enzyme activities to P addition differed considerably across the three ecosystems (forest, grassland, and cropland). P addition significantly decreased the activities of acid phosphatase, phosphodiesterase, and P-acquisition enzymes in forests, but increased the activities of leucine aminopeptidase, carbon acquisition enzymes, and nitrogen acquisition enzymes in grasslands. Sucrase, β-1,4-glucosidase, N-acetyl-β-glucosaminidase, alkaline phosphatase, and carbon acquisition enzymes were significantly stimulated in croplands. In addition, all P addition rates significantly increased the activity of alkaline phosphatase, a P-acquisition enzyme, in croplands. Ca(H2PO4)2 had a greater effect on enzyme activity than other fertilizer types, possibly because Ca2+ can improve the soil structure and enhance microbial metabolic activities. Furthermore, structural equation models revealed that mean annual precipitation and ecosystem type had significant impacts on the response of P-acquisition enzymes, but no factor could directly influence the response of carbon-acquisition enzymes. Our investigation of enzyme activities in response to P addition complements the biogeochemical model and helps forecast soil nutrient dynamics. We propose that rather than focusing on a single factor, P fertilizer addition programs should take into account the effects of mean annual precipitation, ecosystem type, and nutrient interactions, which could also reduce P pollution.