The loss of species diversity is increasingly rapid due to global change. Therefore, understanding how litter species diversity and its composition affect litter decomposition directly or indirectly via mediating soil microbial community has been a hot topic in the field of ecology. In the typical steppe in Hulun Buir, Leymus chinensis is a dominant species in the natural communities, and plants of Artemisia capillaris, Serratula centauroides, and Potentilla bifurca show high abundance in the degraded or recovering communities. In this study, we selected the litter of these four species as study materials, and set up a nested design by nesting litter composition within the factor of litter diversity under the conditions of doubling ambient atmospheric CO₂ concentration. We tried to explore the effects of litter diversity and its composition on litter mass, C, N remaining rate and remaining C/N, and analyze how the litter initial characteristics and the phospholipid fatty acid (PLFA) contents of gram-positive bacteria (G⁺), gram-negative bacteria (G^-), bacteria (B), fungi (F), the soil total microorganisms and F/B impacted the litter decomposition. The results were shown in the following. (1) The factor of litter diversity and its composition had significant effects on mass, C, N remaining rate, and remaining C/N. The litter composition had a significant effect on G⁺ PLFAs content. (2) The result of redundancy analysis showed that litter composition had greater impact than litter diversity on variables involved in litter decomposition and soil microbial community. (3) The result of structural equation model (SEM) indicated that both litter initial lignin content and initial C/N had significant effects on the litter decomposition parameters, and that litter initial lignin content played a more important role. The litter initial lignin content had significant indirect effect on the litter N remaining rate and remaining C/N by changing fungi PLFAs content. These results would provide data support for the prediction of the substance cycling characteristics in the degraded grassland ecosystem and for a better assessment of grassland ecosystem function in the context of global climate change.