The most comprehensive study to date of the effects of size, shape, and flexibility on the translational diffusion of small probe molecules in polymer solutions has been completed by Taylor dispersion, which directly yields D-probe,D- and phosphorescence quenching, which yields k(q), the concentration dependence of which is identical to that of D-probe for appropriate conditions. Diffusion of 16 probes ranging by a factor of 6 in molar volume was investigated using both Taylor dispersion in solutions of up to 400 g/L polystyrene in tetrahydrofuran and phosphorescence quenching in solutions of up to 700 g/L polystyrene in tetrahydrofuran, cyclohexane, and carbon tetrachloride. Results were compared quantitatively to modified Vrentas-Duda free volume theory for ternary solutions to obtain probe jumping unit sizes relative to the solvent, xi(probe,s,) which correlate with probe volume. With the exception of 3,4-hexanedione (a highly flexible and small probe), the PS concentration dependencies of D-probe and k(q) were approximately equal to or greater than that of solvent (0.9 less than or equal to xi(probe,s) less than or equal to 1.75). The data fell into two types of behavior : when xi(probe,s) was plotted against the ratio of probe to solvent molar volume, (V) over tilde(0)/(V) over tilde(0)(s), the vast majority of data fell around a line of slope 0.13, while for two of the probes xi(probe,s) fell near a line of slope unity. Literature data for five probes in several polymer-solvent systems could also be described by these two types of behavior. The former behavior indicates that for most probes the concentration dependence can be described by modified free volume theory, with the understanding that the critical hole free volume for a jump unit for these probes is but a fraction of the probe molar volume. The apparent dichotomy in the probe volume dependence of xi(probe,s) raises the question of whether only two dependencies are possible or whether, by virtue of the probes selected, only these two distinct behaviors are observable. Small effects of flexibility and shape on D-probe for probes with large aspect ratios were also observed and discussed in terms of anisotropic diffusion. A comparison of concentration dependence data with limited temperature dependence data from the literature shows a consistency based on the modified free volume picture. This, along with an understanding of the "bimodal" xi(probe,s) data, indicates that the modified free volume theory for ternary systems forms a reasonably robust picture by which to interpret probe diffusion in polymer solutions.