Dry-sowing and wet-emergence (DSWE) has emerged as a promising water and fertilizer management strategy in arid regions, reducing irrigation water use by similar to 40 % while ensuring uniform seedling emergence. However, research on the mechanisms through which water and fertilizer regimes of DSWE affect cotton yield in southern Xinjiang under DSWE remains limited. This study employed high-throughput 16S rRNA sequencing and structural equation modeling (SEM) to elucidate the pathways linking soil environment, microbial communities, and cotton yield under different water-fertilizer regimes. Results showed that irrigation primarily regulated soil nutrient availability and organic matter accumulation. In contrast, fertilizer type strongly influenced soil NPK contents. Their coupling significantly shaped soil microbial communities, altered nitrogen cycling functions, and ultimately affected cotton yield. Specifically, the W2 quota (517.5 mm) maintained higher soil nutrients across growth stages, and F1 fertilization increased AN, AP, and AK levels by up to 41.34 %. W2 also resulted in the highest SOM, though with reductions of up to 28.10 % in AN and 38.72 % in AP. High-throughput sequencing revealed Proteobacteria (30.47 %), Acidobacteriota (13.62 %), and Chloroflexi (11.99 %) were the dominant phyla. The most complex microbial network was observed under W2, while F2 increased network complexity by 26.44 % compared with F3. Reduced irrigation suppressed chemoheterotrophy and nitrate reduction by 32.36 % and 3.17 %, respectively. SEM indicated that irrigation quotas and fertilizer types regulated key taxa (KD4-96, Sphingomonadales, Xanthomonadales), thereby influencing nitrogen cycling-related functions and yield. PCA and the entropy weight method identified W2F2 as the optimal water-fertilizer combination for efficient cotton production in arid and semi-arid regions.