Physical evidence on the seat belt restraint system is one source of data used by investigators to determine whether or not an occupant was wearing their seat belt during a crash. Evidence of occupant loading on seat belts generated during crash events has been thoroughly researched and is well documented in the literature. However, there is a paucity of data regarding the physical evidence produced when fractured glass is introduced into the restraint system during occupant loading events. The objective of this study is to characterize the physical evidence generated by glass-to-seat belt interaction during low-level impact loading, and compare this evidence with the types of seat belt marks that can be generated inadvertently by accident scene bystanders, emergency responders, and crash investigators. The presence of glass particles in and around the vehicle at the end of a crash event may contribute to the inadvertent generation of physical evidence.Movable side windows composed of tempered safety glass and laminated safety glass were fractured via impactor loading representative of occupant impact. The resulting glass fracture fragments were separated by size using a series of sieves, and the distribution of glass fragments size was quantified.New service-replacement seat belt retractor assemblies (including D-ring, latch plate, anchor, and webbing) were tested using a Seat Belt Load Simulator (SBLS) fixture, which simulates occupant loading by applying a repeatable load pulse to the restraint system. Each retractor assembly was mounted onto the SBLS fixture in a position representative of belt routing when installed in a vehicle. A repeatable lap-shoulder belt stroke pulse, representative of low-level restraint loading and consistent in magnitude and duration with loads produced during rollover, was applied using the SBLS with glass fragments of varying sizes introduced onto the webbing surface adjacent to the D-ring and latch plate surfaces.Additional test series were run to investigate the types of physical evidence generated in the presence of glass under non-crash loading scenarios. These scenarios included the extraction of webbing with glass fragments present adjacent to the D-ring and latch plate, extraction of webbing over glass fragments captured or fixed in a window seal, and the compression of webbing with various sizes of glass fragments interposed between the webbing and a reaction surface.Documentation of each restraint system was performed post-test. Seat belt loading events that occurred in the presence of fractured safety glass produced characteristic markings on the restraint system hardware and webbing. The tests conducted to examine non-crash loading evidence generated in the presence of fractured safety glass revealed markings on the restraint system that differed from those generated in a simulated loading event.