The effects of regular permanent bends on the rotational Brownian motions of flexurally and torsionally deformable DNA models are investigated by Brownian dynamics simulations. Directionality ty of the permanent bends is maintained by anisotropic bending terms in the intersubunit potential. The effects of such bends on the effective hydrodynamic radius for diffusional spinning and on the fluorescence polarization anisotropy (FPA) are determined by comparing simulated results for three different model potentials : (1) a uniform isotropic bending potential with a straight resting geometry (I-model); (2) a uniform isotropic plus anisotropic permanent bending potential, which has the same average projection of a bond vector onto its predecessor as the I-model and which exhibits regular permanent bends (A-model); and (3) a mostly isotropic model, but with a single 90 degrees anisotropic permanent bend at its center(l-model). A 72 base pairs (bp) I-model exhibits practically the same hydrodynamic radius for azimuthal spinning as the corresponding straight cylinder, whereas the A-model exhibits a slightly (3%) larger value over a range of filament lengths. Tile large directional permanent bend of the L-model substantially increases the hydrodynamic radius for diffusional spinning. However, when the anisotropic bending potential is omitted from either the A-or L-models, the permanent bends become nondirectional and have no significant effect on diffusional spinning, which evidently takes place via a uniform speedometer cable rotation. The FPA ofl;he A-model indicates significantly less depolarization than is obtained for the I-model at any given time. The A-model FPA curves can be satisfactorily reproduced by suitably parametrizing the I-model so that (1) the effective hydrodynamic radius for azimuthal rotation is increased to emulate the effect of permanent bends; (2) the dynamic bending rigidity is chosen to emulate the short-time bending amplitude of the A-model; and (3) the rotational diffusion coefficient for end-over-end tumbling is chosen to reflect the total persistence length, including the effect of permanent bends.