Little research has been performed to evaluate the factors that impact seat belt usage for rear seat occupants. Because of the difficulties to collect rear seat belt data in the field, most researches have to rely on crash data, however, has its limit. Seat belt usage passengers not seriously hurt in a crash would typically not be recorded in the police officer filling report. In this study, Rear seat belt models using logistic regression analysis were performed using both FARS data and observational data. The resulting models were then analyzed to determine differences between using crash data and field data in identifying factors associated with the seat belt usage of rear seat occupants. The research showed both similar and differing results between the models produced using FARS data and field data. All the models using the FARS data showed a strong correlation between the back seat passenger’s seat belt usage and the driver’s seat belt usage. The research shows there are differences obtained in the factors that influence seat belt usage using field data and FARS data. The research demonstrates that care must be taken in the use of this data with regard to safety research looking at passenger restraint.