This paper addresses scheduling of quantized power levels (including part load operation and startup/shutdown periods) for a propane powered solid oxide fuel cell (SOFC) hybridized with a lithium-ion battery for a tracked mobile robot. The military requires silent operation and long duration missions, which cannot be met by batteries alone due to low energy density or with combustion engines due to noise. To meet this need we consider an SOFC operated at a few discrete power levels where maximum system efficiency can be achieved. The fuel efficiency decreases during transients and resulting thermal gradients lead to stress and degradation of the stack; therefore switching power levels should be minimized. Excess generated energy is used to charge the battery, but when it’s fully charged the SOFC should be turned off to conserve fuel. However, startup and shutdown phases consume both battery and fuel energy and induce stack degradation, and therefore should be scheduled as infrequently as possible. Simple models of the battery and SOFC are used to evaluate the optimal scheduling strategy using Dynamic Programming. Representative cycles are generated from random sampling of measured power data for specific tasks. Finally a rule-based control strategy is developed and compared with the optimal one, considering battery degradation, fuel efficiency as well as design robustness. The application to military tracked robots for surveillance is considered as an example using power profiles from an instrumented PackBot; however the methodology can be applied broadly to hybrid power systems for transportation which have large turn on/off penalties.