Land use transitions cause reconfigurations of regional landscape patterns which can further change the regional ecosystem service functions and its values, especially in environmentally fragile regions. Firstly, this paper theoretically examines the relationships between land use transitions, landscape pattern evolution and the responses of ecosystem service functions in the Guangxi Zhuang Autonomous Region (Guangxi). Then, it explores the spatio-temporal evolution features of land use transition by using land use change matrices, examines landscape patterns by using the landscape pattern index, and studies ecosystem service value (ESV) by revising the coefficients of ESV per unit area. Finally, focus is placed on the empirical analysis of ESV responses to landscape pattern evolution caused by land use transitions in Guangxi. The results show that: (1) Guangxi has undergone an overall intensity-changing process of land use transition at a moderate rate during 1990–2010 and at a drastic rate during 2010–2018. In general, the area of construction land and waterbodies has increased, while forested land, grassland and farmland have decreased. Landscape fragmentation and heterogeneity are higher in the central area than that in the surrounding areas, while patch aggregation and connectivity show an opposite trend. Forested land patches are highly clustered, while grassland and farmland are fragmented and scattered and construction land patches tend to have aggregated. (2) The total loss of ESV has reached 20.56 billion RMB in Guangxi, and all areas’ single ESVs have decreased to different degrees during the past 28 years. Spatially, the ESV distribution shows a differentiated pattern of low in the central plain and high in the surrounding mountain regions which are mainly dominated by high-value zones. (3) The total ESV has significant positive correlations with the largest patch index (LPI), COHESION and the Aggregation Index (AI), and significant negative correlations with the Number of Patches (NP) and the Shannon Diversity Index (SHDI), while the correlation with the Landscape Shape Index (LSI) is not significant, indicating that the influence on ESV caused by landscape pattern evolution varies greatly. (4) The change of land area and multi-directional shifts among different land use types caused by land use transitions in Guangxi could both lead to the evolution of landscape patterns. Further, ecological service function responded obviously to the landscape pattern evolution in Guangxi, causing significant changes in strengthening or weakening of the ecological service function and its value. This systematic analysis should help coordinate the relationship of regional land use regulation, landscape pattern optimization and ecosystem operation in Guangxi or even China.