This paper proposes a novel two-stage firefly algorithm based on search + adjustment for solving unit commitment problem. The solving process of UC will be broken down into two optimization problems with discrete and continuous variables respectively. The main discrete problem to determine the off/on status is solved by binary encoding firefly algorithm, the continuous sub-problem of economic load variables assignment is solved by improved real-coded firefly algorithm, and check and repair constraints are used to achieve alternative and iterative calculation. Heuristic constraints adjustment, as well as two-stage coding approach is used to achieve the decompose optimization of discrete and continuous variables, and improve the efficiency and accuracy of the algorithm. The algorithm proposed is applied to six different systems and compared with classical algorithms, the results verify the effectiveness and superiority. This work is supported by National Natural Science Foundation of China (No. 51477121).