Landscape performance is an important topic in contemporary landscape architecture. The renewal of small-scale public spaces has a positive effect on community vitality and the sustainable development of landscape resources. To improve the quality of small-scale public spaces, a clear understanding of landscape performance and its correlative factors is necessary. The structural equation model (SEM) provides a way of objectively evaluating the influence of public space on landscape performance, which could be useful in the design decision-making process. This study develops a methodological framework that implements SEM to assess landscape performance and provides a case study of typical small-scale public spaces represented by subway entrances in Nanjing. Five latent factors and 11 observed factors were selected to construct a landscape performance assessment model. Various methods, including investigation, image recognition, and modeling, are exploited to analyze the characteristics of the subway entrances quantitatively. An SEM-based calculation is conducted to obtain the correlation coefficient of latent factors and the explanatory degree of observed factors. Statistical analysis indicates that the green space and traffic capacity affect the landscape performance of the subway station entrances. Moreover, the fluctuation of any latent factor may cause the decay or enhancement of related factors. Therefore, we suggest that design strategies for subway station entrances should balance traffic, visibility, plants, and facilities. More importantly, this study works as a technical support and reference point for future empirical research to obtain an in-depth understanding of the ideal landscape construction forms in small-scale public spaces making it possible to predict the level of landscape performance when a new study area is given.