Accelerations of the head are the likely cause of concussion injury, but identifying the specific etiology of concussion has been difficult due to the lack of a valid animal or computer model. Contact sports, in which concussions are a rising health care concern, offer a unique research laboratory environment. However, measuring head acceleration in the field has many challenges including the need for large population sampling because of the relatively low incidence of concussions. We report a novel approach for calculating linear acceleration that can be incorporated into a head-mounted system for on-field use during contact sports. The advantages of this approach include the use of single-axis linear accelerometers, which reduce costs, and a nonorthogonal arrangement of the accelerometers, which simplifies the design criteria for a head-mounted and helmet compatible system. The purpose of this study was to describe the algorithm and evaluate its accuracy for measuring linear acceleration magnitude and impact location using computer simulation and experimental tests with various accelerometer configurations. A 10% error in magnitude and a 10 deg error in impact location were achieved using as few as six single-axis accelerometers mounted on a hemispherical headform.