Abstract: 【Objective】It is of great significance to clarify the farmers’ nutrient input situation for the realization of stable yield, high yield, and high nutrient use efficiency in dryland wheat production.【Method】A 5-yr long farm survey of 1 261 farmers was carried out to analyze and evaluate their fertilizer applications and the fertilizer reduction potential in Weibei dryland, based on the nutrient requirement determined by the corresponding wheat grain yields and sustainable development of dryland wheat production.【Result】Obtained results showed the farmers’ wheat yields ranged from 750 to 9 000 kg hm^-2, with the average of 4 243 kg·hm^-2, and they were allocated into five groups as: very low (＜2 640 kg·hm^-2), low (2 640-3 780 kg·hm^-2), moderate (3 780-4 920 kg·hm^-2), high (4 920-6 060 kg·hm^-2) and very high (＞6 060 kg·hm^-2), respectively, accounting for 22.0%, 22.2%, 19.3%, 22.8% and 13.6% of the total. Farmers’ nitrogen (N) application rates ranged from 33 to 454 kg N·hm^-2 with an average of 188 kg N·hm^-2, phosphorus (P) ranged from 0 to 435 kg P₂O₅·hm^-2 with an average of 125 kg P₂O₅·hm^-2, and potassium ranged from 0 to 201 kg K₂O·hm^-2 with an average of 19 kg K₂O·hm^-2. However, farmers’ yields showed no significant correlations with the N, P, and K rates, respectively. With the increase of grain yield levels, the proportion of N over application farmers decreased from 97.8% in the very low yield group to 18.0% in very high group, but that of N deficient application farmers increased from 0.7% to 45.9%, correspondingly. Similar to N, the proportion of P over application farmers decreased from 99.3% in very low yield group to 70.9% in very high yield group, and this means P over application was practiced by more than 70.0% of farmers in each yield group. Different from N and P, K deficient application was practiced by more than 60.0% of farmers in each yield group. Therefore, for N, farmers in very low and low yield groups were recommended to reduce 24-144 kg N·hm^-2, 28%-73% from their high or very high N application rates, and farmers in moderate, high and very high yield groups were recommended to reduce 50-181 kg N·hm^-2, 26%-51% of their high or very high N rates and add 38-134 kg N·hm^-2, 41%-345% more to the low or very low N rates. For P, farmers in different yield groups should reduce 7-31 kg P₂O₅·hm^-2, 23%-33% from the high P rates, and reduce 85-118 kg P₂O₅·hm^-2, 61%-85% from the very high P rates. For K, farmers with no or very low K input in different yield groups were suggested to use 13-50 kg K₂O·hm^-2, and add 7-18 kg K₂O·hm^-2, 35%-78% for those with low K rates. 【Conclusion】 Compared with the conventional method, which adopted an uniform fertilization rate as the criterion to evaluate the famers’ fertilizer application with variable yields, the present work proposed a yield based approach. This approach is proved to be suitable for the small scale household farming in China, and enable to objectively and accurately understand the arbitrary and over application of fertilizer, and to provide a scientific basis for the effective regulation of farmers’ fertilizer application.

Key words: farmer, winter wheat, grain yield, fertilizer application rate, fertilizer recommendation