Lodging is an important agronomic trait related to crop yield and is easily susceptible to environmental influences. In this study, a recombinant inbred line population from soybean (Glycine max (L.) Merr.) Hefeng 25 × Dongnong L28 including 109 lines was used to identify quantitative trait loci (QTLs) related to soybean lodging. Seven QTLs were identified in the three environments (Harbin in 2017, 2018 and 2019), and these could explain 2.21–20.17% of the phenotypic variation. Among these QTLs, qLDG-I-1 (Chr20_24146101–Chr20_24297321) was stable for multiple environments. A residue heterozygous line, which was heterozygous at the qLDG-I-1 locus, was used to verify qLDG-I-1, and the results showed that this QTL could significantly improve lodging resistance of soybean. Meanwhile, 13 pairs of epistatic QTLs were detected, which could explain 3.26–18.24% of the phenotypic variation. QTL × environment interaction mapping was also used, and it detected 31 QTLs, which could explain 1.61–7.94% of the phenotypic variation. In total, 122 pairs of epistatic QTLs were detected, and they could explain 5.39–27.81% of the phenotypic variation. Additionally, candidate genes related to soybean lodging in the qLDG-I-1 interval were predicted, and Glyma.20g068000 was mined as a candidate gene based on quantitative real-time PCR analysis. The QTLs and candidate genes identified in this study are of great significance to position cloning, and could accelerate the progress of breeding resistance to lodging in soybean.