In this study, we aim to improve the poor convergence behavior of high-frequency electromagnetic field problems and to accelerate the computational time by introducing multiple-precision arithmetic into a parallel finite element method for high-frequency electromagnetic fields. The developed code is implemented in ADVENTURE_FULLWAVE , an open-source parallel finite element analysis software for high-frequency electromagnetic fields based on the hierarchical domain decomposition [4] framework. Multiple-precision arithmetic is realized using MPFR (Multiple Precision Floating-Point Reliable Library). MPFR is a software library implemented in the C programming language that enables floating-point arithmetic with very high (arbitrary) precision. As a preliminary study, we applied our self-developed COCG method incorporating MPFR to the complex symmetric matrix qc2534, which is registered in the Matrix Market / UF Sparse Matrix Collection and has a condition number of 1.819206×10^5. As a result, for qc2534, the COCG method using double-precision arithmetic did not converge; however, by applying MPFR, convergence was achieved, and it was found that the number of iterations decreases as the mantissa precision is increased. However, with respect to the computational time per iteration, the calculation time is found to be up to approximately 100 times longer than that of double-precision arithmetic. Even in domain decomposition methods, the computational time may increase; however, by applying MPFR only to the interface problem or by solving certain parts of the computation with higher precision, the number of iterations may be reduced, potentially leading to a reduction in the overall computational time. In this congress, we present the implementation of MPFR for domain decomposition methods, including MPI communications, as well as the results of a parameter study on the partial introduction of multiple-precision arithmetic.