Recently, Webster et al. have shown that there is excess reddening in radio-selected quasars relative to the optically selected population. If this reddening is universal to all quasars, then the optical surveys that depend on the UV-excess selection criteria, say, would be seriously incomplete. Using data compiled from the literature and various statistical tests, we show that there is no significant correlation between B - K colors and other reddening indicators in the case of optically selected quasars. The distribution of emission-line equivalent widths and line ratios of radio-loud and radio-quiet quasars in the optically selected sample are consistent with their being drawn from the same parent population. Our results suggest that there is an intrinsic spread in B - K color of ~2 mag in optically selected bright quasars due to effects other than dust.Our model calculations suggest that the required amount of reddening cannot be produced by dust in the intervening damped Lyα absorbers. We estimate, for different extinction curves, the optical depth of dust intrinsic to quasars that is required to produce the observed spread in optical-to-near-IR colors. Results of photoionization models suggest that, for a wide range of ionization parameter and metallicity, the gas associated with dust will produce Lyman limit as well as saturated heavy-element absorption at the redshift of the quasar. Observation of such associated absorption in very red quasars will confirm reddening due to dust intrinsic to the quasar, but the data presently available suggest that the ~2 mag spread in B - K color in the sample is due to effects other than dust extinction. We present marginal evidence for the aspect dependence of optical-to-near-IR colors in radio-loud quasars. This points to beaming as a possible reason for the reddening, but further investigation using a homogeneous sample is necessary before a conclusion can be reached.