Forensic toolmark examiners compare marks between those observed on an item/surface and those made by a reference implement, such as a particular tool or weapon, to provide an opinion of the likelihood of common origin. It is widely accepted that such comparison opinions need to be underpinned by empirical research, and this study aimed to add to the knowledge base relied upon when developing and comparing saw marks in bone, a substrate encountered in body dismemberment cases. Porcine bones were used as a human proxy; they were either fresh with residual soft tissue and bodily fluids present (‘wet’) to replicate dismembered bones shortly post-mortem, or processed to remove soft tissue and moisture content (‘dry’) to represent cases of dismemberment after an extended period of decomposition and exposure. The bones were cut using one implement of each of five classes: hand saw, mitre saw, reciprocating saw, oscillating saw, and serrated knife. They were cut, either completely through (except for serrated knife), giving two surfaces per cut to examine, or to a depth up to 3 mm (false starts). Five replicates per combination of bone condition, saw, and cut type gave 130 bone samples. These were then cleaned and cast using Isomark Silicone Polymer Compound or Mikrosil, giving 260 cast samples. All bone and cast samples were photographed, examined for various class characteristic markers, and specific markers measured. No significant differences between Isomark and Mikrosil casts were observed when compared side-by-side, demonstrating suitability of both materials for casting of saw marks on bone. Although saw marks presented more class characteristic markers on dry than wet bones, calculations of tooth distances and measurements of kerf width (KW) from marks did not significantly differ between bone conditions, with exception of the reciprocating saw that produced false start marks with significantly larger minimum KW on wet than dry samples. Further analysis supported that tooth distances on marks made by hand and oscillating saws are sufficiently accurate for the determination of saw teeth per inch (TPI). However, one tooth distance on marks made by reciprocating saws did not accurately represent TPI. Finally, examination of presence or absence of class characteristic markers on each saw mark demonstrated consistent variation between saw classes. These results enabled the development of exclusion-based decision trees, and a reference database (available on request), for use by toolmark examiners in their evaluation of saw types based on class characteristic markers observed in cut bone.