The RAS procedure is widely used to update national and regional inputoutput tables and international trade tables and to construct regional tables from national ones. Special problems, however, have been encountered when the procedure is used to adjust interregional trade tables. In this paper, the special properties of interregional trade tables that increase the likelihood of nonconvergence of the RAS procedure are discussed, and two linear programming methods of solving infeasible RAS problems are provided. First, a closed linear programming approach, which enables exogenous information to override the purely mechanical solution of infeasible RAS problems, is presented. Finally, the open linear programming approach is applied successfully to adjust US. interregional trade data that had previously failed to converge using the RAS procedure.The major purpose of this paper is to present a new way in which interregional trade flows can be adjusted to marginal control totals through the combined use of two estimation techniques: the RAS and linear programming procedures. There is a large and growing literature on the use of nonsurvey techniques to estimate input-output tables, but only a few attempts have been made to use nonsurvey techniques for the adjustment or estimation of interregional trade flows. Of the numerous nonsurvey techniques that have been developed, the RAS procedure is one of the most widely used for input-output tables. It is so named because it involves the estimation of an unknown matrix from a known matrix, A, subject to row and column control totals R and S, respectively. An extensive survey of the use of the procedure on input-output data is provided in an earlier paper by the three authors [Polenske, Crown, and Mohr (1986)l.In the first section of this paper, a brief introduction is given to the RAS procedure, and a review is provided of the four known attempts to use the RAS procedure for adjustments of trade-flow tables. In the second section, the theoretical and empirical issues related to the present study are provided. In the third section, two new theoretical proofs concerning a linear programming procedure for determining critical cells in the tables are set forth. In the fourth section, applications of the proposed RAS and linear programming procedures to the multiregional input-output (MRIO) data are discussed. In the final section, major conclusions of the research are presented, with emphasis on the similarities and differences in the use of the RAS procedure for interregional trade tables versus its