A Monte Carlo (MC) simulation is used to evaluate the performance of an optical differential phase-shift keying (DPSK) communication system impaired by in-band crosstalk. The MC simulation results are used to study the validity range of an analytical work proposed in the literature addressing the same problem. It is shown that the performance estimates obtained using the analytical work become incorrect, whenever the effect of inter-symbol interference (ISI) is enhanced, for example, for optical filter bandwidths below twice the DPSK signal bandwidth. When ISI is negligible, a slight discrepancy is observed between the performance estimates of both methods, because the two formalisms consider different models for the random phase noise.