A large research program was carried out to evaluate the effectiveness of different heavy vehicle rear underrun guards. Based on the test results, a recommended underrun guard specification was developed to prevent passenger injuries in such accidents. Data obtained from a series of static and full-scale crash tests performed at the Transport Canada Research Center was used in the development of the specification. Four different underrun guard designs were used during the full-scale crash tests. Three different vehicle models representing sub-compact, compact vehicles and light trucks and vans traveling at 48, 56 and 65 km/h relative speeds were used to impact underrun guards head on. The first 10 of the crash tests guard was attached to a simulated trailer while the last one performed using an actual truck. The results obtained from these tests show that a guard built to the minimum requirements of U.S. FMVSS 223 may not be strong enough to prevent passenger compartment intrusion especially in compact and sub-compact automobiles. This paper documents the tests performed, the results obtained and their analyses, and conclusions derived from these analyses. A recommended specification is prepared on the minimum performance that should be required for an effective underrun guard.