Abstract: A daptively modified characteristic basis function method (AMCBFM) based on QR factorization with a dual modified Gram-Schmidt (dual-MGS) algorithm called QR-AMCBFM is proposed. In this hybrid method, the object geometry is partitioned into distinct blocks, which are divided into near and far groups depending on distances. The primary basis functions are derived firstly. In order to get the high-level basis functions, QR factorization algorithm based on dual-MGS is applied to decompose the mutual coupling matrixes to decide that the rest interactions arising in farther blocks are ignored or not. Subsequently, the high-level characteristic basis functions and the coefficient of current are calculated. Results show that the QR-AMCBFM can solve the problems very efficiently.