AbstractSoil testing laboratories require predictive equations that provide accurate fertilizer recommendations to improve economic return to producers and reduce the risk of environmental pollution. Snow management by cereal trap strips and fertilizer N management involving rates, placement, and timing of application were studied in a 9‐yr experiment on a Swinton Loam (Aridic Haploboroll) in southwestern Saskatchewan. Spring wheat (Triticum aestivum L.) was grown annually using zero‐tillage management. Stepwise regression with backward elimination was used to develop the following relationship: Y = −781 + 3.11 WU + 6.88 FN − 0.053 FN2 − 0.33 SN2 + (0.23 SN × WU) − (0.25 SN × FN) + (0.035 FN × WU) + (0.00083 SN × FN × WU) + (1.60 Yr × SN) − (0.53 Yr × FN) R2 = 0.91, P=0.001, n = 1248), where Y = grain yield (kg ha−l), WU = estimated water use (mm), SN = available soil N (kg ha−1), FN = rate of fertilizer N (kg ha−1), and Yr = number of years of study. Water use was available spring soil water in the 0‐ to 1.2‐m depth plus 1 May to 31 July precipitation, and SN was NO3‐N in 0‐ to 0.6‐m depth, measured in the fall. The regression accurately predicted yields of wheat for an 8‐yr zero‐tillage study carried out on similar soil nearby. Fertilizer N placement (seed placed, deep band, broadcast) influenced grain yields more than timing, but in 5 of 9 yr neither factor was significant. Response to fertilizer N decreased with years, presumably because the available N supplying power of the soil improved under no‐till management with adequate fertilization. Efficiency of available N use [grain wt/(SN + FN)] varied from 4 to 41 kg grain kg−1 N, being directly related to WU and generally to SN, but inversely to FN rate.