In the development of new electronic systems the planning of reliability growth tests has become both more critical and more difficult as available testing budgets have diminished. Previously, system designers were able to plan and implement relatively lengthy reliability growth test plans to assist in the development of reliable systems. A new method is presented to allocate subsystem reliability growth test time in order to maximize the system mean time between failure (MTBF) or system reliability when designers are confronted with limited testing resources. For certain problems, the algorithm yields the same results as competing approaches to the problem but with significantly fewer required iterations. More significantly, the new algorithm applies to a larger problem domain compared to analogous algorithms. Much more realistic formulations of the problem can now be solved optimally. The algorithm is based on objective function gradient information projected onto a feasible region that consists of candidate test plans given the testing budget constraint. The algorithm is demonstrated on several examples with superior results.