Data from the National Accident Sampling System (NASS) are used to construct linear regression models with prior recorded accidents as the dependent variable. Independent variables investigated include driver age, miles of driving experience, prior moving violation convictions of several types, and driver training history. The distribution of the NASS data were found to be generally consistent with previously reported research. The results of univariate and multivariate analyses demonstrate that heavy truck drivers form a discreet subset of all accident-involved drivers, whose accident history is more reliably predicted by a more parsimonious model as compared to drivers of other vehicle types.