The cropland definition was adopted from the Joint Experiment of Crop Assessment and Monitoring (JECAM) network, which defines the annual cropland from a remote sensing perspective as a piece of land of a minimum 0.25 ha meters that is sowed/planted and harvested at least once within the 12 months after the sowing/planting date (Waldner et al., 2016). Considering that cropland might be fallow in a single year, the dataset will use the remote sensing data of three consecutive years (2019-2021) to detect the cropland, and in turn ensure its proper classification. A field will be considered as cropland when it is detected as cropland at least once over the three-year period. This product was generated using all available Landsat series and Sentinel-2 satellite imageries during 2019-2021. Due to the difference in crop calendar in the northern and southern hemispheres, there are minor variations in the acquisition dates of remote sensing data used in different agro-ecological zones. Most regions of the northern hemisphere used data from February 2, 2019 to February 1, 2022, while data in many regions in the southern hemisphere were generated using remote sensing data from November 1, 2018 to October 31, 2021. Cropland maps in a few regions were extracted based on satellite imageries from late August 2018 to mid-August 2021 due to unique crop phenology. Accuracy was first assessed over eight regions with each covering an area of 10º×10º with different agricultural practices. The overall accuracy of the extracted cropland is 90.5% over the eight regions. The assessment was further expanded to global scale, showing the overall accuracy of global high-resolution cropland in 2020 at 89.4%.