A statistical optimization based on experimental work was conducted to consider ultimate tensile strength (UTS) and elongation of dissimilar joints between AA5454 and AA7075 by friction stir weld (FSW). The goal of this work is to develop a comparative study of the optimization of FSW parameters using different orthogonal arrays, i.e., L12 and L16. Four parameters correlated to softening and forging requirements (rotational speed, traverse speed, tilt angle, and plunge depth), one parameter associated with the location of base metal in the dissimilar joint, and two parameters related to an FSW tool (pin profile and Dshoulder/dpin ratio) were considered and arranged in the employed arrays. Moreover, the investigation explored the microstructure and fractography of dissimilar joints and base metals by using optical and scanning electron microscopes. The results showed that the L16OA is more accurate than L12OA for the optimization of seven parameters due to the small statistical errors. For UTS, the errors range from 0.78 to 24% for L16OA and from 27.23 to 44.14% for L12OA. For elongation, the errors run from 11 to 12.9% for L16OA and from 33.77 to 49.73% for L12OA. The accuracies of generated models range from 50 to 99.5% for L16OA and range from 30.7 to 94.9% for L12OA. Tightening the levels (narrow domain) is the main reason for switching some optimum levels between both arrays. The highest UTS obtained is 221 MPa based on the optimum levels attained from L16OA, and the highest elongation is 12.83% according to the optimum levels acquired from L12OA. Despite the deficiency of effective intermixing, the study revealed that FSW acceptably could assemble joints between AA5454 and AA7075, presenting the proficiency of FSW with welding dissimilar aluminum alloys.