A detailed study at the Missouri University Research Reactor indicates that limitations in the energy balance methodology, using the Monte Carlo N-Particle transport code (MCNP) and the Evaluated Nuclear Data Files (ENDF), affect the accuracy of predicting important parameters for reactor physics studies. In the case of fuel conversion, key parameters such as flux and power level cannot be measured until the converted reactor is operating. Therefore, predictions with well-known uncertainties are essential for an effective conversion. However, due to inherent energy balance problems in the isotopic heating evaluations for materials within various fuel matrices, in particular the U-10Mo monolithic fuel, the values for the predicted parameters could vary more than previously estimated. In particular, the total recoverable energy per fission, which directly affects the calculated flux for a given power level, appears to be underestimated by MCNP's energy balance method. Therefore, an alternative methodology for predicting the total recoverable energy of a system was investigated. Results for the proposed low-enriched uranium U-10Mo configuration show that there is a 3.02-MeV difference between the total recoverable energy per fission from this work and that from the MCNP predictions. A similar comparison for the present highly enriched uranium UAlx configuration shows a difference of 1.24 MeV.