Aiming at the nonlinear projection decomposition problem in the image reconstruction by dual-energy computed tomography (DECT), a fast projection decomposition algorithm based on isotransmission line fitting is proposed. A linearly approximated isotransmission line fitting model is established for high and low energy projections. Based on the Taylor series expansion, the coarse-grain forward calculation and fine-grain reverse calculation are used to get the analytical solution with high accuracy, and the isotransmission line equations are generated. The nonlinear solution problem is transformed into the problem to solve intersections of high and low energy projection isotransmission lines, and thus the fast decomposition of dual energy projections is realized. The experimental results show that the proposed method avoids the nonlinear iteration and the projection matching, reduces noise fluctuation in projection decomposition, and improves the projection decomposition rate significantly. Compared to the projection matching method, the proposed method improves the decomposition rate by about two orders of magnitude.