Summary. The agronomic effectiveness of a partially acidulated phosphate rock (PAPR), produced by the 50% acidulation of North Carolina phosphate rock with sulfuric acid, was compared over 4 growing seasons with triple superphosphate (TSP) and the highly reactive North Carolina phosphate rock at 22 permanent pasture sites in the National Reactive Phosphate Rock Project. The performance of PAPR as a phosphorus (P) fertiliser for permanent pasture was determined by calculating the substitution value of TSP for PAPR at 50% of the maximum yield response for TSP from the fitted annual dry matter response curves. PAPR performance varied both between sites, and between years at individual sites. Annual yield responses with PAPR were larger than those with TSP at 1 high rainfall site where water-soluble P from TSP was thought to leach from the root zone. PAPR was superior to TSP at another site and generally similar in effectiveness to TSP at 4 sites with light-textured, low or medium P-sorbing soils with a moderate annual rainfall (500–750 mm). The mean substitution value over the 4 years for these sites was >0.9. PAPR performance at other sites where highly reactive phosphate rocks were effective in the short or medium term was variable: there were equivalent yield responses to TSP in some years but much smaller yield responses in other years. PAPR performed very poorly in a third group of sites where the soil had a high to very high P sorption capacity or where there was a very high demand for fertiliser P due to large legume responses on a P-deficient soil. Although generally inferior to TSP, the PAPR was more effective than North Carolina phosphate rock at the majority of sites during the 4-year study.