The Genetic Algorithm (GA) requires randomized initial population so as to give requisite results. The paper proposes the use of Cellular Automata (CA) to generate the initial population for GA. The numbers generated by clubbing these two processes will make a better set of random numbers as compared to existing methodologies. The quality of the random numbers generated has been tested by Gap Test and Karl Pearson coefficient of correlation test. So as to ensure the goodness of random numbers coefficient of auto correlation has also been calculated.