The excitation temperature T_01 derived from the relative intensities of the J = 0 (para) and J = 1 (ortho) rotational levels of H2 has been assumed to be an accurate measure of the kinetic temperature in interstellar environments. In diffuse molecular clouds, the average value of T_01 is ~70 K. However, the excitation temperature T(H3+) derived from the (J,K) = (1,1) (para) and (1,0) (ortho) rotational levels of H3+ has been observed to be ~30 K in the same types of environments. In this work, we present observations of H3+ in three additional diffuse cloud sight lines for which H2 measurements are available, showing that in 4 of 5 cases T_01 and T(H3+) are discrepant. We then examine the thermalization mechanisms for the ortho:para ratios of H3+ and H2, concluding that indeed T_01 is an accurate measure of the cloud kinetic temperature, while the ortho:para ratio of H3+ need not be thermal. By constructing a steady-state chemical model taking into account the nuclear-spindependence of reactions involving H3+, we show that the ortho:para ratio of H3+ in diffuse molecular clouds is likely governed by a competition between dissociative recombination with electrons and thermalization via reactive collisions with H2.