Pork bellies (n = 198) were scanned with dual energy X-ray absorptiometry (DXA). Visible and near-infrared reflectance (Vis-NIR) spectra were collected from the lean (latissimus dorsi), subcutaneous fat and intermuscular fat layers. Belly-flop angle and subjective belly scores were collected as measures of pork belly softness. Vis-NIR spectra from a single fat layer could explain between 72.7 and 81.1% of the variation in pork belly softness (43.6–72.4% in validation set). The combination of the lean and subcutaneous layers improved the calibration model fit to 79.7–99.9% (66.3–71.5% in validation set). The DXA estimates explained 62.3% of variation in pork belly softness (65.2% in validation set). Results indicated that DXA and NIR technologies could potentially be utilized for pork belly softness sorting in the pork industry.