The estimation of postmortem submersion interval (PMSI) has always been an important scientific issue to be addressed in drowning cases. Traditional methods, such as corpse temperature analysis and the assessment of corpse surface corruption, have limitations and cannot meet the need for accurate estimation of the time of death in the mid to late stages. Biogenic amines, as small molecules produced by protein degradation after death, have a certain regularity in relation to PMSI. To further explore the possibility of utilizing polyamines to estimate PMSI, this experiment constructed a rat cadaver model in both laboratory constant-temperature water and natural water bodies. Furthermore, cadaverine and putrescine in the liver and skeletal muscle were detected at different PMSI using gas chromatography-mass spectrometry (GC-MS). Through statistical analysis, we have constructed eight sets of mathematical models for polyamines-PMSI estimation, and determined the applicable time range through derivative analysis. After evaluation the models, the error rate in inferring PMSI using the fitted equations was less than 30 % within 242 h. The models established in this study could accurately infer PMSI in the mid to late stages of the postmortem period, providing a feasible approach for the drowning forensic issue.