The effectiveness of safety belts in preventing fatalities to drivers and right front passengers is estimated by applying the double pair comparison method to 1974 or later model year cars coded in the Fatal Accident Reporting System. The method focuses on "subject" occupants (drivers or right front passengers) and "other" occupants (any except the subject occupant). Fatality risks to belted and unbelted subject occupants are compared using the other occupant to estimate exposure. In this study, drivers and right front passengers are subject occupants; choosing other occupants differing in age, seating positions, and belt use, generated 46 essentially independent estimates of safety belt effectiveness. The weighted average and standard error of these is (41 +/- 4)%. This finding agrees with the 40%-50% range reported in a recent major review and synthesis by the National Highway Traffic Safety Administration. Combining this with the present determination gives (43 +/- 3)%; that is, if all presently unbelted drivers and right front passengers were to use the provided three point lap/shoulder belt, but not otherwise change their behavior, fatalities to this group would decline by (43 +/- 3)%.