The Big Bend Regional Aerosol and Visibility (BRAVO) study was an intensive air quality study designed to understand the causes of haze in Big Bend National Park. Daily speciated fine aerosols were measured from July through October 1999 at 37 sites located mostly in Texas. In support of BRAVO, two chemical transport models (CTMs) were used to apportion particulate sulfate at Big Bend and other sites in Texas to sources in the eastern and western United States, Texas, Mexico, and the Carb¿n I and II coal-fired power plants, located 225 km southeast of Big Bend in Mexico. Analysis of the CTM source attribution results and comparison to results from receptor models revealed systematic biases. To reduce the multiplicative biases, a hybrid source apportionment model, based on inverse modeling, was developed that adjusted the initial CTM source contributions so the modeled sulfate concentrations optimally fit the measured data, resulting in refined daily source contributions. The method was tested using synthetic data and successfully reduced source attribution biases. The refined sulfate source attribution results reduced the initial eastern U.S. contribution to Big Bend, averaged over the BRAVO study period, from ~40% to ~30%, while Mexico's contribution increased from 24--32% to ~40%. The contribution from the Carb¿n facility increased from ~14% to over 20%. Contributions from Texas and the western United States changed little, with final contributions of ~16% and 5--9%, respectively. The increase in Mexico's contribution is consistent with more recent SO2 emissions estimates that indicate that the BRAVO Mexican SO2 emissions were underestimated. Source attribution results for other monitoring sites in west Texas were similar to results at Big Bend. In eastern Texas, the eastern United States accounted for up to 70% of the measured sulfate, with Texas contributing ~20--30%.