An efficient prediction method for the elution profile was used to separate aromatic compounds such as benzene, toluene, chlorobenzene, o-xylene, and 1,2-dichlorobenzene by RP-HPLC. The retention factor and bandwidth were predicted under linear-gradient condition with the three retention models. The elution profiles were calculated based on the linear and quadratic equations of retention factor, lnk = lnk(w) + S, lnk = L + M phi + N phi(2), k = A+B/phi, where phi was the vol.% of methanol. The elution profiles were calculated by the Gaussian distribution with obtained retention factor and bandwidth. Two kinds of experiments were performed; one is the isocratic runs to estimate the coefficients of three retention models, and the other is the linear gradient runs that were carried out with same initial mobile phase composition (water/methanol = 96/4, vol.%), two final mobile phase compositions (water/methanol = 24/76 and 40/60, vol.%), and three gradient times (20, 40, and 60min). The predicted elution profiles by the three retention models and new prediction method have good agreement with experimental data in the employed gradient conditions. The minimum average errors of calculated and experimental results of aromatic compounds were lower than 3.5% by Bi-poly equation.