Under the interaction between natural ecosystems and human interferences, farmland extends to multi-functions such as production, ecological, social, and cultural functions. Despite the exponential increase in research on the multi-functional evaluation of farmland in recent years, little study has been conducted at fine spatial and long-time scales. Furthermore, the existing quantitative analyses of multifunctional synergies and trade-offs in farmland mainly consider static spatial patterns and neglect dynamic information. Selecting the Chinese coastal province of Zhejiang as the study area, this study thus evaluated the spatio-temporal patterns of farmland functions from 2000 to 2020 at the county scale and introduced the trade-off/synergy degree (TSD) model to quantify the intensity of the relationships among functions. The results showed that farmland functional values and their relationships were significantly heterogeneous in spatial and temporal distribution. In addition to social function, the other functions all exhibited an increasing trend. Furthermore, strong correlations were mainly observed between production, ecological and cultural functions. Ultimately, five farmland zones were determined by the k-means clustering algorithm and considering both functional values and their relationships, and targeted suggestions applicable to each zone were put forward in this study. This study contributes to the utilization and planning of farmland and its surrounding land, especially to the improvement of the policy of returning farmland to forests.