Fencing is one of the effective techniques to restore degraded steppes, which has been applied in the steppes of north China for many years. Due to different vegetation and soil conditions in lightly, moderately, and heavily degraded steppes, it is still unclear whether fencing has a consistent effect on plant community and soil properties of the steppes along the different degradations. We conducted a six-year fencing experiment in lightly, moderately, and heavily degraded desert steppes in the Inner Mongolia, China and measured the plant community and soil properties in the growing season of 2021. The results showed that fencing significantly increased height, cover, and aboveground biomass at population (i.e., Stipa breviflora and Cleistogenes songorica) and community levels (P<0.05), which drove the vegetation restoration of degraded desert steppe from multiple organization levels. Fencing generally increased plant diversity of the lightly and moderately degraded desert steppes but decreased that in the heavily degraded desert steppe. The recovering efficiency of community height, cover, and aboveground biomass in the heavily degraded desert steppe was significantly higher than those in the lightly and moderately degraded desert steppes (P<0.05), indicating that fencing had more effective for vegetation restoration in the heavily degraded desert steppe. Moreover, except for the lightly degraded desert steppe, fencing significantly decreased soil total carbon, total nitrogen, total phosphorus, available nitrogen, and available phosphorus concentrations of moderately and heavily degraded desert steppes (P<0.05), but had no remarkable effect on soil moisture along the three degraded desert steppes. This result showed that fencing had lag effects on soil physical and chemical properties compared with plant community. Our study provides theoretical guidance for evaluating the effects of fencing on desert steppe and scientific basis for implementing fencing policies accurately in the degraded desert steppe ecosystems.