This paper presents a new approach to test generation from extended finite state machines using genetic algorithms, by proposing a new fitness function for path data generation. The fitness function that guides the search is crucial for the success of a genetic algorithm; an improvement in the fitness function will reduce the duration of the generation process and increase the success chances of the search algorithm. The paper performs a comparison between the newly proposed fitness function and the most widely used function in the literature. The experimental results show that, for more complex paths, that can be logically decomposed into independent sub-paths, the new function outperforms the previously proposed function and the difference is statistically significant.