A methodology is presented for developing indirect estimates of cross-border sales of motor fuels when data on price differentials are not available. The extent of motor fuel sales occurring on Native American tribal reservations is not precisely known, as no existing data source provides a comprehensive survey of on-reservation sales. Significant unreported sales could create bias in the attribution process used by FHWA to apportion each state's share of federal tax receipts. A methodology is developed to estimate the magnitude of fuel sales on Native American tribal reservations. As significant tax rate differentials may exist, prices on and off reservation may vary considerably. For the most part, these price differentials are not observed, and doing so would require a comprehensive (and expensive) survey. The model described does not rely on knowing the actual price differential but rather generates indirect estimates of sales to nonresidents of the tribal areas. This is done by comparing predicted sales in counties adjacent to Native American tribal reservations with service stations (generated by the model) with actual sales. The resulting estimates suggest that for counties adjacent to tribal reservations with significant fuel sales, cross-border sales may account for as much as 8% of total consumption. These results are robust to numerous model specifications.