We examine carefully the accuracy of stochastic rotational dynamics (SRD) simulations for isolated polymer chains in a solvent, where SRD incorporates hydrodynamic interaction (HI) through momentum exchange (collisions) between polymer beads and solvent beads, both of which are assigned mass. We show that the main error is due to the inertial effect that finite bead mass has on polymer hydrodynamics. We find that the inertial effect is negligible when R-g, the radius of gyration of the polymer chain is much larger than L-M, the distance over which bead inertia is lost due to collisions with solvent. For moderate HI, good agreement is found between the rotational relaxation time simulated by SRD with that from normal-mode analysis and from Brownian dynamics (BD) simulations, even for short five-bead chains. For dominant HI, for short chains, we can minimize the inertial effect by varying the ratio of polymer to solvent bead mass. For long chains (R-g >> L-M) SRD and BD relaxation times agree, but are larger than those from normal-mode analysis due to neglect of fluctuating HI in the latter. We also find that, using the same parameters, the SRD method can reproduce the BD results obtained by Jendrejack et al. for a k-DNA chain in viscosified water. (C) 2013 The Society of Rheology.