A genetic algorithm (GA) with the condensed encoding and the improved fitness function is used to solve Japanese puzzles. In this study, the condensed encoding can make sure that the chromosome is a feasible solution in rows for Japanese puzzles. In the reconstruction process of a Japanese nonogram, the numbers in the left column are used as encoding conditions, and the numbers in the top row with the improved fitness function are employed to evaluate the reconstruction result. From the computational experiments, the GA is effectively applied to solve a Japanese puzzle.