The optimization of the stand structures of low-efficiency forests is often required to sustainably achieve the full range of ecological benefits from soil and water conservation in forests. This study was conducted in the Caijiachuan watershed of the Loess Plateau in western Shanxi, China, and explored the practical, and optimal configuration of stand structures in a Robinia pseudoacacia forest. Accordingly, one hundred and ninety-five standard plots (20 × 20 m²), including six stand densities (~500, ~1000, ~1500, ~2000, ~2500 and ~3000 plants·hm⁻²), were established. Twenty functional indicators and ten structural indicators were monitored and investigated from June to September each year from 2017 to 2018. The results suggested that the forest was divided into three levels of low efficiency, which were mild (II), moderate (III), and severe (IV), based on the soil and water conservation ecological benefits index (SWBI). The trends of the polarization of the values of the stand structure indicators appeared in forests II, III and IV, and this phenomenon became increasingly obvious with increasing levels of low efficiency. Analysis of the influence factors indicated that the climate of this region exhibited a warming and drying trend, the soil moisture content presented a decreasing trend, and the vegetation growth rate slowed. The stand density (SD), tree height (TH), canopy density (CD) and the leaf area index (LAI) were the main controllable factors that influenced the low SWBI of the R. pseudoacacia forest. The results of the optimal configuration of the stand structure suggested that the SD should be controlled between 1000 and 1600 plants·hm⁻², the TH should be controlled between 7 and 9 m, the CD should be controlled between 0.6 and 0.8, and the LAI should be controlled between 2.5 and 5. In comparison to the SWBI of the existing stand structure configuration, the optimal configuration is expected to improve the SWBI by approximately 86%, 200% and 600% in R. pseudoacacia forests I, II and III, respectively. These results will provide a more scientific and complete reference for the optimization of the stand structures of low-efficiency R. pseudoacacia forests.