The dynamic second-order hyperpolarizability from the third-harmonic generation experiment, 〈γ〉(−3ω;ω;ω;ω), of 59 solution-phase organic molecules has been theoretically calculated and compared with available experimental data. The DFT/6-311+G(2d,p)//(CAM-)B3LYP/6-31G(2df,p) level of theory with the polarizable continuum model was employed to evaluate the static second-order hyperpolarizability. The theoretical static values were then corrected to dynamic estimates of 〈γ〉(−3ω;ω;ω;ω) by employing a two-level approximation (Chen et al. in J Chem Phys 101(7):5860–5864, 1994). Calculated results showed that for the property of interest, the range-separated hybrids provided more accurate and consistent estimate than the global hybrids and the CAM-B3LYP optimized geometry delivered more accurate estimate than the B3LYP optimized geometry.