A 14-year field trial was conducted on a calcareous soil to evaluate the effects of continuous rice (Oryza sativa L.) cropping and fertilization on crop yield, soil fertility and phosphorus fractions. The application of N and P enhanced rice yields, while K had no yield-increasing effect because of large available soil K resource. The soil organic carbon remained at a stable level except for a decrease in the unfertilized treatment after 1988. Total N did not show clear changes with time or between treatments despite some fluctuation. The available soil K trend after 1985 as well as the comparison between treatments indicated clearly decreasing available soil K, particularly in the NP treatment. Available soil P significantly decreased in the P-omitted treatments, but remained at a stable level in the P-applied treatments with time. Of the total inorganic P (P i ), Ca phosphates (Ca-P) were the dominant P i forms, accounting for 69-71%, followed by Fe phosphates (Fe-P), P occluded within Fe oxides (O-P) and then Al phosphates (Al-P). Of the Ca-P fractions, Ca 2 -P [CaHPO 4 .nH 2 O], Ca 8 -P [Ca 8 H 2 (PO 4 ) 6 .nH 2 O] and Ca 1 0 -P [Ca 1 0 (PO 4 ) 6 .(OH) 2 ] accounted for 3.1-6.2, 5.8-6.4 and 87-91%, respectively. The NK treatment had significant Ca 2 -P depletion with time compared with the NPK treatment, and the similar trend was observed for Ca 8 -P despite decreasing Ca 8 -P with time for the two treatments. In the NK treatment, Ca 2 -P and Ca 8 -P had a good correlation with Olsen-P. The significant changes with time were not observed for Fe-P, O-P and Ca 1 0 -P. The P application caused a weak accumulation of Fe-P, O-P and Ca 1 0 -P, but had no significant effect on total P i over time. The results suggest that (i) P application is indispensable to maintain high yields of rice under N application and (ii) there was a substantial P release from Ca 2 -P and Ca 8 -P linked to P uptake by crops. Hence, soil indigenous P supply, P transformation and the yield responses to fertilizer P application must be synthetically considered to optimize fertilization strategies for irrigated rice production on the calcareous soil at the experimental site.