The present study aims to determine the set of optimum process parameters in order to minimize spiking during bead-on-plate welding of ETP copper plates using electron beam. Accelerating voltage, beam current, welding speed, focus distance, amplitude and frequency of oscillations were considered as the input design variables. The spiking was characterized in terms of average penetration and standard deviation of penetration. Statistical regression analysis was conducted to correlate the spiking phenomena with the input parameters of electron beam welding. The reasons behind the spiking phenomenon and the associated porosity defects were also explained. This unconstrained optimization problem was solved using genetic algorithm, particle swarm optimization, and desirability function approach in order to maximize weld penetration and minimize standard deviation of penetration. The obtained optimized results were verified through real experiments and a good agreement between them was achieved. Genetic algorithm was found to perform slightly better than other approaches.