In order to overcome difficulties of time-domain airborne electromagnetic (AEM) data inversion and interpretation resulted from abrupt variations in traditional single-station inversions, we develop a stable and efficient quasi-3D spatially constrained inversion (SCI) scheme.We first design two procedures of fixed and variable layer-thickness in our quasi-3D inversion. Then, weimposed in the objective functional constraints to model parameters in directions respectively along the survey line and perpendicular to it and use L-BFGS algorithm to minimize the objective functional for the optimal solution of the quasi-3D problem.The resultsfrom our quasi-3D SCI inversion and from single-station inversions onboth syntheticand survey data demonstrate that the algorithm of spatial constraints is more effective in suppressing the noise and delivering smoother and more continuous results in the underground.