Qiandongnan is the largest and best-preserved Miao settlement area in China, holding significant ethnic cultural heritage. Traditional settlements form an essential part of this heritage. Studying their spatial characteristics and influencing factors is crucial for the sustainable development and protection of cultural heritage in this region. By comprehensively utilizing ArcGIS spatial analysis, boundary morphology index, spatial syntax, and geographic detector methods, the spatial characteristics of settlements were deconstructed from the perspective of regional pattern and case feature analysis, and their influencing factors were explored. The results indicate that the spatial distribution of traditional Miao settlements in Qiandongnan is characterized by significant agglomeration and hierarchy. The highest nuclear density is at the intersection of Leishan, Taijiang and Jianhe. The overall spatial pattern shows a "dense in the southwest and central-south, sparse in the northeast" distribution. Constrained by natural geography, the settlements are mainly distributed in the Qingshui River and Duliu River valleys at altitudes of 500 ~ 1 000 m, with undulations of 10 ~ 20 m, gradients of 2° ~ 5°, and sunny slopes of 90° ~ 270°. The settlements' spatial structure exhibits a "clustered" distribution with finger-like external boundaries, and the center shows differentiated traffic flow within and at the edges of the settlements. The geodetector study reveals that Miao traditional settlements are regional spatial carriers of a natural-economic-social complex system. The natural geographic environment fundamentally shapes spatial patterns, the social environment guides and controls internal spatial organization and evolution, and economic development decisively influences spatial development and protection. The study enhances the understanding of this complexity, which is vital for appreciating Miao culture, developing strategies for protecting and developing this cultural heritage, and implementing rural revitalization strategies.