The objective of this study was to determine if a novel scoring-based model for histological quantification of decomposed human livers could improve the precision of post-mortem interval (PMI) estimation for bodies from an indoor setting. The hepatic decomposition score (HDS) system created consists of five liver scores (HDS markers): cell nuclei and cell structure of hepatocytes, bile ducts, portal triad, and architecture. A total of 236 forensic autopsy cases were divided into a training dataset (n = 158) and a validation dataset (n = 78). All cases were also scored using the total body score (TBS) method. We specified a stochastic relationship between the log-transformed accumulated degree-days (log10ADD) and the taphonomic findings, using a multivariate regression model to compute the likelihood function. Three models were applied, based on (i) five HDS markers, (ii) three partial body scores (head, trunk, limbs), or (iii) a combination of the two. The predicted log10ADD was compared with the true log10ADD for each case. The fitted models performed equally well in the training dataset and the validation dataset. The model comprising both scoring methods had somewhat better precision than either method separately. Our results indicated that the HDS system was statistically robust. Combining the HDS markers with the partial body scores resulted in a better representation of the decomposition process and might improve PMI estimation of decomposed human remains.