A major challenge for motorist safety and animal welfare is designing measures to avoid animal-vehicle collisions. Many mitigation measures have been proposed and implemented, but few are informed by data on animal behavior. One type of behavioral data that may be helpful in the design of mitigation is animal and driver behavior around the moment of collision. We obtained a collection of over 400 publically posted videos of vehicle-animal collisions from dashboard cameras. We quantified both animal and driver behavior, for example where the animal collided with the vehicle and a driver’s steering reactions just before and after collision. We found that that smaller cervids (deer) move at a high speed perpendicular to the roadway such that drivers have little chance of avoiding collision, even if the driver is alert and driving safely. In contrast, moose cross more slowly and with more varied crossing trajectories, which increases the time for alert drivers to react effectively. Differences in species behavior when entering a roadway should be considered during mitigation measure design. For example, mitigation measures based on motion detection may be less effective at reducing collisions with deer that dart into the roadway at the approach of a vehicle.