Remote sensing technology using satellite images has become more common to quantify the pasture biomass. However, there are still problems in predicting pasture yield in Hokkaido due to local factors (e.g., rapid growth in early spring, deterioration of vegetation). Thus, the purpose of this study was to analyze the relationship between pasture yield and satellite image-based vegetation indices in Hokkaido. The study area is 13 pasture fields in Kamishihoro, Hokkaido. A total of 9 field surveys were conducted from April to August 2021. Pasture was harvested twice during the experimental period. We sampled pasture at one point for each field by the quadrat method. The Sentinel-2 L2A satellite imagery products were used, and Normalized Difference Vegetation Index (NDVI) and Enhanced Vegetation Index (EVI) were calculated. We investigated the correlation between yield and NDVI or EVI. There was a positive correlation between yield and NDVI, EVI during the first growing period, but there was no correlation during the second growing period. Our study suggested that predicting the second pasture yield is more difficult because of differences in the regrowth rate depending on the grass species and more heterogeneity, such as weed appearance and uneven heading production in pasture fields.