Subsidence of lunar mascon maria, impact basins partly filled with mare basalt and sites of prominent positive gravity anomalies, typically led to the formation of concentric graben (arcuate rilles) around the flanks of the basin, while compressive features (mare ridges) formed in interior regions. Although previous numerical models of the response of the lunar lithosphere to mascon loading predict that an annulus of strike-slip faulting should also have formed around mascon maria, no such faults have been observed. This strike-slip faulting paradox, however, arises from an oversimplification of the earlier models. Viscoelastic finite element models of lunar mascon basins that include the effects of lunar curvature, heterogeneous crustal strength, initial stress conditions, and multistage load histories show that the width of a predicted annulus of strike-slip faulting may be small. The use of Anderson's criterion for predicting fault styles may also overpredict the width of strike-slip faulting. A faulting-style criterion that takes into account transitional faulting, in which both strike-slip and dip-slip components are present, predicts zones of pure strike-slip faulting that are about half of the width predicted by the Anderson criterion. Furthermore, strike-slip faulting should be observed only in regions in which flexural stresses are sufficient to induce rock failure. However, since stress patterns consistent with strike-slip faulting around mascon loads represent a transition between compressional and extensional provinces, differential stresses tend to be low in these regions and for at least part of this region are not sufficient to induce rock failure. A mix of concentric and radial thrust faulting is observed in some mascon maria, at odds with previous models that predict only radial orientations away from the basin center. This apparent discrepancy may be partly explained by the multistage emplacement of mare basalt units, a scenario that leads to a stress pattern where concentric and radial orientations of thrust faults are equally preferred. Detailed models of the Serenitatis basin indicate a 25-km-thick lunar lithosphere at the time of rille formation and a 75-km-thick lithosphere at the time of late-stage mare ridge formation. The extent of observed mare ridges and the inferred cessation of rille formation around Serenitatis prior to the time of emplacement of the youngest mare basalt units is consistent with the superposition of a global horizontal compressive stress field generated by the cooling and contraction of the lunar interior with the local stresses associated with lithospheric loading. ¿ 2001 American Geophysical Union