Quantifying the energy associated with vehicle damage is the basis of common methods used to reconstruct car crashes. This study sought to characterize the relationship between crush and energy for the front and rear surfaces of a passenger car. Nine stationary barrier crash tests and one aligned car-to-car test were conducted using several cars of the same model with impact speeds ranging from 4.3 to 15.2 m/s generating as much as 0.47 m of crush. The results revealed a linear speed-crush relationship for front and rear car surfaces and a restitution coefficient that decreased from a maximum of 0.33 at low speed to a relatively constant value of 0.15 for crush levels above 0.2 m. Crush coefficients derived from the crash tests were compared to the coefficients from three other sources: i) default values from the CRASH3 computer program, ii) values from a published database and iii) values derived from an assumed damage threshold value and an NHTSA high-speed crash test. Crush coefficients generated using these methods did not reflect the difference between front and rear stiffness properties indicated by the crash test data. Only calculations using crush coefficients derived directly from the tests and including the effects of restitution accurately predicted the actual barrier impact speeds and speed changes. Based on our results, assuming the same front and rear damage thresholds when calculating crush coefficients may not be appropriate. Considering a range of damage threshold values from 2.2 m/s to 4.5 m/s would have, at least for the car used in this study, provided a range of predicted impact speeds that bracketed the actual values. Ignoring the effects of restitution, or assuming a restitution coefficient of zero, in the derivation and application of crush coefficients will tend to under-estimate the actual speed change.