Recent complexities in interconnected power system operation call for more efficient and fast approaches for operational problems including network‐constrained unit commitment (NCUC). In this study, an algorithm called period‐elimination Benders decomposition (PEBD) is proposed to enhance the computational efficiency of AC NCUC problems. After linearising the AC NCUC, two subproblems (SPs) are defined to ensure power balance at buses as well as constraints of branches and inter‐regional transactions. In the classical Benders decomposition (BD), all time periods are solved at all iterations up to convergence. However, in the proposed PEBD, a time period that satisfies a SP in an iteration is eliminated from the solution procedure of that SP at next iterations. After both SPs are satisfied for all time periods, the solution is checked for interrelation constraints and a few more iterations may be needed to finalise the solution. The efficiency and accuracy of the proposed PEBD method are verified by examining it on two test systems. It is found that the PEBD greatly enhances the speed of NCUC, especially in larger power systems due to eliminating unnecessary calculations for satisfied time periods.