Reliable normalization is essential for accurate quantitative real-time PCR (qRT-PCR) analysis, yet suitable reference genes have not been systematically evaluated in Mugilogobius chulae, an emerging marine experimental fish used in physiological and environmental research. In this study, 17 candidate reference genes were evaluated in five tissues (brain, gill, gonad, intestine, and liver) of sexually mature male and female M. chulae under solvent control (DMSO), bisphenol A (BPA), cadmium (Cd), and sulfamethazine (SMX) treatment conditions. Gene-expression stability was assessed using the comparative ΔCt method, NormFinder, geNorm, BestKeeper, and the integrated RefFinder ranking. The results showed that reference-gene stability was strongly tissue-specific and analysis-dependent. The integrated RefFinder analysis identified eif3h, rpl7, rps4x, stau1, and ef1y as the most stable genes in the pooled dataset, whereas ube2, hprt1l, and aldob were the least stable. geNorm analysis indicated that two reference genes were sufficient for brain, gill, intestine, and liver, whereas four were required for the gonad and six for cross-tissue comparisons. These findings provide the first systematic basis for reference-gene selection in M. chulae and establish an important methodological foundation for future qRT-PCR studies in this species, particularly in research on endocrine disruption, reproductive physiology, and marine ecotoxicology.