This paper reports recent work undertaken by the Oxford Road Accident Group to improve the utility of the CRASH 2 (Calspan Reconstruction of Accident Speeds on the Highway) program. Although CRASH 2 is being used extensively in the U.S.A. for accident reconstruction, its use in Europe is limited. The accident environment in Europe is sufficiently different from that in the United States that it has been suggested that the program could be of limited use under these conditions. Accordingly, to provide reliable figures as to the utility of the program under European conditions a representative sample of accidents has been reconstructed using CRASH 2: the sample consisted of 200 accidents investigated on-scene and 200 accidents investigated on a 48 hour follow-up basis. Results are presented which give the proportion of accidents that could be successfully CRASHed together with the reasons for not running CRASH.To improve the reconstruction accuracy of CRASH 2 for European sized vehicles, the stiffness coefficients used in the damage analysis routine have been reassessed for the minicar, subcompact and compact car categories. The data used for the update consisted of 42 frontal impacts performed in the U.K. The new coefficients were derived using a statistical optimisation technique to fit the damage equation to the test data. The new coefficients show a significant improvement in the accuracy of individual reconstructions. Provision for vehicles to be yawing at impact has also been added to the program.