Estimating postmortem interval (PMI) of surface found skeletal remains is challenging. This novel study used UV-Vis-NIR spectroscopy to scan soil collected from cadaver decomposition islands (CDIs) ranging from 15-to 963-d postmortem and control soils. A decomposition product spectra model (DPS model) was constructed by deducting the control soil spectra from the CDI soil spectra for the estimation of postmortem indices: PMI (d), ADD 4 , ADD 10 , and ADD 20. The DPS model (n = 55) was calibrated and subjected to a full cross-validation. Calibration R 2 and RPD for the DPS model ranged from 0.97 to 0.99 and from 6.1 to 9.9, respectively, for the four postmortem interval indices. Validation R 2 and RPD for the DPS model ranged from 0.73 to 0.80 and from 1.9 to 2.2, respectively. The DPS model estimated postmortem intervals for three test CDIs in a clay soil under perennial grassland (test set 1; n = 3) and six CDIs in a sandy soil under a loblolly pine forest (test set 2; n = 6). Test set 1 had PMI prediction ranges from −69 to −117 days, −796 to +832 ADD 4 , +552 to +2672 ADD 10 , and −478 to −20 ADD 20 of observed PMI. Test set 2 PMI prediction ranged from −198 to −65 days, −9923 to +2629 ADD 4 , −6724 to +1321 ADD 10 , and −2850 to +540 ADD 20 of observed PMI. Test set 2 had poor predictions for two CDIs, for all measures of postmortem indices resulting in discussion of sampling depth, effect of body mass index (BMI), and scavenging.