Robust optimization of interplanetary trajectories is difficult becaus
e the domain is typically multimodal and discontinuous. Genetic algori
thms are often successful in such domains. Therefore, a genetic algori
thm has been linked to a patched-conic mission analysis code to assess
its performance for interplanetary problems. It is compared with grid
search techniques, which are commonly used in the initial exploration
of interplanetary concepts. This new method out-performs grid search
in all test problems considered to date. The genetic algorithm's perfo
rmance superiority is shown to increase with the number of design vari
ables. When a sharing function is included, the genetic algorithm is a
ble to identify several local minima. Performance was influenced by th
e topology of the search space: large infeasible regions increased the
difficulty of the problem. The genetic algorithm is able to explore d
ifferent mission concepts simultaneously, allowing simpler problem for
mulation for the user.