Environmental conditions contribute to a large percentage of wheat yield variability. This phenomenon is particularly true in rainfed environments and non-responsive soils to N. However, the effect of P application on wheat is unknown in the absence of N fertilizer application. This study was conducted from 2012 to 2019 in permanent beds established in 2005. Treatments were arranged in a split-plot design and consisted of superimposing three P treatments (foliar, banded and broadcast application) plus a check (0P) within each one of four preceding N treatments (applied from 2005 to 2009). Foliar P generally showed a greater response than granular P treatments even though the soil tests high P (>30 mg/kg). Precipitation estimated for two different growth intervals explained through regression procedures the Years' effect. Seasonal precipitation (224–407 mm) explained variation of relative yield, N harvest index (NHI) and P agronomic efficiency (AE). Reproductive stage precipitation (48–210 mm) explained soil N supply. In dry years, foliar P application improved predicted relative yield 14% and AE 155 kg grain/kg P compared to granular P treatments. Similarly, soil N supply increased 15 kg/ha in dry moisture conditions during the reproductive stage. The NHI consistently improved over the crop seasons. This improvement was relatively larger for 0 kg N/ha. On average, NHI increased from about 0.57 to 0.72%. Normalized difference vegetation index (NDVI) readings at the booting growth stage were negatively associated with NHI. Foliar P in this non-responsive soil to N showed the potential to replace granular P sources. However, the omission of granular P needs to be further studied to estimate the long-term effect on the soil P test.