Based on the theory of human-environment system, a theoretical analytical framework for the urban underused land was built. Taking 106 counties in Shanxi Province as the study unit, the spatial autocorrelation method and Geodetector method were employed, the spatial differentiation pattern of urban underused land in Shanxi Province was analyzed, also the leading factors for spatial differentiation of urban underused land in county territories in Shanxi Province were quantitatively detected, so as to explore the interaction mechanism between the urban underused land in county territories and spatial factors. The results showed that with the analytical framework of “inefficiency-nature-economy-society” built, the acting path between urban underused land and spatial factors as well as their pattern of manifestation analyzed.It was found that, on the whole, two types of factors, respectively “human” and “earth”, and three dimensions, respectively, the natural dimension, the economical dimension and the social dimension, were there, among which the natural factor played a basic role in the spatial differentiation of underused land, while the economic factor played a decisive role, and the social factor had a strengthening and amplifying effect. Totally 89.09% of the urban underused land in county territories in Shanxi Province was less than 265.86hm2, showing the features of low-value agglomeration and weak homogeneous agglomeration;in spatial distribution, the urban underused land in highly-developed county territories was more than that in not so highly-developed county territories, and the urban underused land in counties which were in a flat area was more than that in counties which were in a mountainous and hilly area. The size of permanent resident population, GDP, degree of external transportation convenience, regional average GDP and the number of industrial enterprises above designated size were the leading factors for spatial differentiation of urban underused land in Shanxi Province, though the driving forces of each factor were obviously different at the county level. Among various factor interactions, the non-linear enhancement type was of majority, supplemented by the double-factor enhancement type. Since the size of permanent resident population had the strongest influence on the factor interactions, the active effect of the population factor got highlighted.