Human activities affect CO₂ budget strongly in heterogeneous agricultural landscape. Clarifying the dynamics of CO₂ flux and its environmental control is important for making carbon sequestration increase and emission reduction strategies in heterogeneous agricultural landscape. In this study, CO₂ flux was observed with tall-tower eddy covariance continuously in a typical heterogeneous agricultural landscape of Yangtze River Delta in order to investigate CO₂ source and sink characteristics and its environmental controls. The results showed that the diurnal variation of CO₂ flux presented that CO₂ was absorbed in the daytime and was emitted at night in spring, summer and autumn. The diurnal variation range of CO₂ flux was -0.15-0.14, -0.60-0.28, -0.21-0.19 mg m^-2 s^-1 in the three seasons, respectively. In winter, CO₂ was released throughout the day, and the diurnal variation range was 0-0.14 mg m^-2 s^-1. The agricultural landscape absorbed CO₂ from June to November, and released CO₂ in the rest months of each year. The CO₂ absorption and emission peaks appeared in August and November of each year, with the multi-year mean values of -0.14 and 0.08 mg m^-2 s^-1, respectively. The agricultural landscape was carbon source for the atmosphere. The CO₂ emission increased from 2019 to 2021. The multiyear averaged annual total CO₂ flux was (142.73±99.01) gC m^-2 a^-1. The photosynthetic photon flux density (PPFD), soil temperature at 10 cm depth (T_s), and saturated vapor pressure deficit (VPD) were the key environmental factors affecting the CO₂ flux on half-hour scale. On the daily scale, PPFD and T_s were the main control factors. On the monthly scale, T_s was the most important influencing factor, while precipitation (Pre) and normalized difference vegetation index (NDVI) also affected the CO₂ flux. The anthropogenic CO₂ emission within 15 km of the flux tower obtained by the Open-source Data Inventory for Anthropogenic CO₂ was (429.7±30.01) gC m^-2 a^-1, which offset the carbon sink effect of natural ecosystems such as paddy fields and forests in the agricultural area.