Soil-test correlation and calibration, a useful tool for fertilizer recommendations, has been little used in West Africa. Soils from a long-term fertility experiment have been used to study the relationship between rice yields and soil extractable phosphorus (P) with Bray 1 and Olsen methods. The Cate and Nelson graphical method was used for critical limits of soil P determination. The critical limits of soil extractable P at 95% relative grain yield were 9 mg P for the Bray 1P and 17 mg P kg21 for Olsen P. The Olsen P was more correlated (r ¼ 0.63) with rice grain yields than Bray P (r ¼ 0.50), but a strong correlation (r ¼ 0.92) was also observed between the values of the two methods. Results indicate that at levels less than these critical levels of extractable P, P fertilizers should be applied to increase rice yields.