Land degradation and soil deterioration are key environmental problems in dry-hot valleys of southwestern China, where vegetation deterioration and soil erosion have reached critical levels in these fragile ecological zones. Restoration programs in these valleys depend on scientific information about which tree species are best for rehabilitation programs, as well as the responses of edaphic constraints to these tree species. To date, knowledge about the amelioration effects of different plantation types on degraded soils in dry-hot valleys is scarce. In this study, we evaluate five monospecific tree plantations (Leucaena leucocephala, Albizia kalkora, Acacia auriculiformis, Azadirachta indica, and Eucalyptus camaldulensis) and one self-repair treatment established to restore a degraded dry-hot valley. We conducted this study to improve our understanding of the effects of vegetation restoration on the soil amelioration process and to provide a theoretical basis for the selection of tree species and restoration practices for rehabilitation programs in dry-hot valleys. Soil characteristics associated with each of the six regeneration treatments were investigated four times over the last 22 years of vegetation restoration. Results show that both regeneration treatment type and elapsed time have significant effects on the soil characteristics of the degraded soils during restoration. Although most properties of the soils associated with the six regeneration treatments improved substantially over the last 22 years, soil fertility and particularly the soil structure were still less well developed than those of undisturbed soils at equivalent sites. After 22 years of vegetation restoration, soil microbial and chemical properties of the degraded soils had ameliorated to >90% and >60%, respectively, compared with those of undisturbed soils; whereas soil physical properties ameliorated <30% in all six treatments. During the entire restoration period, soil physical properties were only enhanced by 3.0%-20.0%, which is significantly less than that observed for soil microbial and chemical properties. The percent of soil amelioration success via self-repair mechanisms was 63.6%, which was higher than the amelioration success for plantings of Albizia kalkora (54.3%), Azadirachta indica (54.9%), and E. camaldulensis (53.2%) but was less than that recorded for plantings of L. leucocephala (68.2%) and Acacia auriculiformis (67.3%). Thus, the different tree species clearly affected soil amelioration processes to different extents. During the process of soil amelioration, soil microbial properties were regenerated first, followed by improvements in chemical properties, while physical properties improved only slightly over the 22-year study period. This study also shows that manual restoration of vegetation (i.e., afforestation) did not always accelerate soil amelioration relative to natural restoration (i.e., self-repair) in dry-hot valleys. We conclude that L. leucocephala and Acacia auriculiformis are the most suitable species as pioneer trees for soil amelioration of degraded soils in these regions. Cost-free self-repair of degraded soils is also a practical option for soil amelioration in regions where afforestation is difficult to carry out. The results of this study need to be further evaluated using a wider range of vegetation types and other degraded valley-type savannas with different soil characteristics to test the general applicability of our conclusions.