ABSTRACT Our objective was to evaluate the agronomic and industrial sugarcane parameters under different P application methods in two cycles of sugarcane cultivated in tropical soils with different mineralogic composition. The experiment was carried out in sandy (PVAd1), sandy-clay (PAdx) and sandy-clay-loam (PVAd2) soils during two sugarcane cycles (plant-cane and ratoon-cane). We tested the localized and non-localized application methods of P with and without fractionation of the P rate. We evaluated extraction, allocation, and efficiency use of P, as well as the growth, yield, and technological quality of the sugarcane. The non-localized application provided greater P extraction, regardless of the soil mineralogy, however showed a greater residual effect in ratoon-cane in the PVAd1. The non-localized application increased the P content in the juice in plant-cane in the PVAd2 but reduced the recoverable sugar. The agronomic efficiency index showed that the localized application is suitable for PVAd2 while the non-localized application is recommended for PVAd1 in plant-cane. Thus, the non-localized application and the P fractionation showed better agronomic and industrial parameters in tropical soils, especially on the sandy despite the localized application being the traditional method.