Objectives Based on multi-year, multi-site field experiments and on-farm survey data in the Weibei dryland region, this study aimed to assess the changes in winter wheat 1000-kernel weight (TKW) and explore the key factors affecting TKW, and then provide a theoretical basis for increasing winter wheat yield.

Methods Field experiments and on-farm surveys were conducted in nine counties of the Weibei dryland region, Shaanxi Province from 2016 to 2023. Wheat plants and soil samples were collected at maturity for analysis. Based on the mean TKW, all samples were classified into high (H), medium (M), and low (L) groups using a ±10% deviation threshold. Relationship between TKW and yield components, nutrient uptake, and climate conditions were subsequently analyzed.

Results The average TKW was 49.6 g, with 62.8% of the samples ranging from 42 to 54 g. TKW was positively correlated with grain yield and kernel number per spike (Y=0.0014x+39.8482, R²=0.17, P<0.01; Y=0.2288x+42.7303, R²=0.04, P<0.01, respectively), negatively correlated with spike number (Y=−0.0084x+53.5725, R²=0.03, P<0.05). Compared with the H group, grain yield in the M and L groups decreased by 11.6% and 33.4%, respectively; kernel number per spike decreased by 7.8% and 10.4%, whereas spike numbers increased by12.9% and 14.1%, respectively. Nitrogen application rates in the H and M groups were significantly higher than that in the L group, and the phosphorus application rates in the M group were significantly higher than that in the L group. Significant differences were found in nitrogen content in different organs among the groups. Compared with the H group, grain nitrogen content in the M and L groups decreased by 7.1% and 13.0%, respectively. Moreover, compared with the L group, aboveground P uptake in the M and H groups increased by17.1% and 19.6% respectively, and aboveground K uptake increased by40.6% and 44.7%, respectively. The TKW had a significant positive linear correlation with the precipitation during the period from sowing to regreening and the summer fallow period, respectively (Y=0.017x+40.7348, R²=0.06, P<0.01; Y=0.0278x+34.7865, R²=0.13, P<0.01), whereas it had a significant negative linear correlation with the precipitation during the period from anthesis to maturity (Y=−0.029x+46.4653, R²=0.03, P<0.01) and the growing degree-days during the grain-filling period (Y=−0.0434x+68.8940, R²=0.04, P<0.01) , respectively. Significant differences in both TKW and grain yield were observed among wheat cultivars. Random Forest analysis indicated that precipitation from sowing to the regreening, cultivar, and summer fallow precipitation were the primary drivers affecting TKW, collectively accounting for 26.0% of the total variable importance. Among fertilization rates, plant nutrient indices, and soil nutrient indices, the most influential factors were identified as N application rate, grain K content, and soil available K, respectively.

Conclusions Winter wheat TKW was significantly associated with grain yield in the Weibei dryland region. Precipitation from sowing to the regreening, cultivar, and summer fallow precipitation were identified as the primary driving factors affecting TKW. In production, it is recommended to enhance soil moisture conservation during the early growth stages, together with rational N and P fertilization and appropriately increasing K fertilizer application, to enhance wheat TKW in this area and thereby improve food security.