Abstract: Based on a long-term fertilization experiment in Guizhou Province, we explored the relationships between the soil available phosphorus (Olsen-P), soil apparent P balance and P application rate in order to quantify the best application rate of P fertilizer in yellow upland soil of southwestern China. Moreover, the response curve of crop yield to soil Olsen-P was fitted by Mitscherlich equation to determine the critical content of Olsen-P for crop yield. The results showed that the long-term application of P fertilizer could significantly increase the content of soil Olsen-P, and the increasing rates of Olsen-P across different treatments could be mainly explained by the application rate of P fertilizer. Under no-P treatment, the soil P content was in a deficient state for each year, while it displayed a surplus state in the treatments with P fertilizer, and the crop P uptake and P accumulation were found the highest under MNPK treatment. In contrast to single application of chemical fertilizer treatment (NPK), the combined application of organic fertilizer and chemical fertilizer (1/4 M+3/4 NPK, 1/2 M+1/2 NPK) could enhance crop P uptake and improve accumulative P use efficiency. The soil apparent P balance was significantly (P<0.05) correlated with soil Olsen-P. With average P accumulation of 100 kg·hm^-2, the soil Olsen-P increased by 16.4, 13.0 , 21.4 , and 5.6 mg·kg^-1 in the treatments of MNPK, 1/4 M+3/4 NPK, 1/2 M+1/2 NPK, and NPK, respectively. The result showed that combined application of organic fertilizer and chemical fertilizer could effectively increase the soil Olsen-P content, and the critical value of soil Olsen-P was 22.4 mg·kg^-1 in yellow upland soil of southwestern China. The soil P balance and Olsen-P content were significantly (P<0.01) correlated with the annual P application rate. When the amount of average P application was 33.3 kg P·hm^-2·a^-1, the budgets of soil P balance remained stable, and the application rate of P fertilizer corresponding to the critical value of soil Olsen-P for crop yield was 45.9 kg P·hm^-2·a^-1. The content of soil Oslen P was mainly affected by the P fertilizer input amount. When the average P application rate was 45.9 kg P·hm^-2·a^-1, higher crop yield and P fertilizer efficiency would be achieved. When the average P application rate was greater than 45.9 kg P·hm^-2·a^-1, crop yield showed no response to P fertilizer input, but resulted in a large amount of P surplus in soil, thereby increasing the environmental risk of soil P loss. The long-term application of manure resulted in a higher increase of Olsen-P than the single chemical P ferti-lizer.