The genetic algorithm (GA), an optimization technique based on evolution, suffers often from a phenomenon called the premature convergence. That is, the system often loses the diversity of the population at an early stage of searching. In this paper, the authors propose a novel method called the ThermoDynamical Genetic Algorithm (TDGA), which adopts concepts of the temperature and entropy suggested from the simulated annealing (SA) to maintain the diversity of the population. Further, the computational complexity of TDGA is evaluated, and comparative study of TDGA with the Simple GA is carried out taking a knapsack problem as an example.