Objective To determine the effect of dose calculation accuracy on inverse linear energy transfer (LET) optimization for intensity-modulated proton therapy, and to determine whether adding more beams would improve the plan robustness to different dose calculation engines. Methods Two sets of intensity-modulated proton therapy plans using two, four, six, and nine beams were created for 10 prostate cancer patients: one set was optimized with dose constraints (DoseOpt) using the pencil beam (PB) algorithm, and the other set was optimized with additional LET constraints (LETOpt) using the Monte Carlo (MC) algorithm. Dose distributions of DoseOpt plans were then recalculated using the MC algorithm, and the LETOpt plans were recalculated using the PB algorithm. Dosimetric indices of targets and critical organs were compared between the PB and MC algorithms for both sets of plans. Results For DoseOpt plans, dose differences between the PB and MC algorithms were minimal. However, the maximum dose differences in LETOpt plans were 11.11% and 15.85% in the dose covering 98% and 2% (D2) of the clinical target volume, respectively. Furthermore, the dose to 1 cc of the bladder differed by 11.42 Gy (relative biological effectiveness). Adding more beams reduced the discrepancy in target coverage, but the errors in D2 of the structure were increased with the number of beams. Conclusion High modulation of LET requires high dose calculation accuracy during the optimization and final dose calculation in the inverse treatment planning for intensity-modulated proton therapy, and adding more beams did not improve the plan robustness to different dose calculation algorithms.