To generalize the characterization of particle-size cut performance for aerodynamic cyclones, a data-driven modeling approach using two varied correlating strategy Psi(0.5)= C(c)(1/2)d(p0.5)/D and Stk(0.5)=(C-c.rho(p)d(p0.5)(2)v(i))/(18 mu D) was proposed. This approach correlated the global influencing parameters including cyclone dimensions, operating conditions and multiphase properties which have not ever completely included before. The proposed models integrated the external geometrical dimensions (cyclone inlet area (ab/D) and vortex finder diameter D-e/D) into a annular Reynolds number Re-a=rho gQ(D-D-e)/[mu(ab)], and used a special body dimensionless parameter D-B/D derived from an equivalent volume method to characterize effect of the cyclone body dimensions: vortex finder length S/D, cyclone height H/D, cylinder height h/D and particle outlet diameter B/D. Results showed that the proposed Psi(0.5)-based correlation improves the predictive ability and generalization performance compared to the other corresponding theoretical and regression models. Furthermore, the influence of cyclone dimensions and operating parameters on particle cutoff size was quantitatively addressed based on the proposed model. The result may provide a reference for performance assessment, design improvement and global optimization of aerodynamic cyclones applied both industrial process and aerosol sampling.