Abstract. The size structure and biomass of a phytoplankton community during the spring bloom period can affect the energy use of higher-trophic-level organisms through the predator–prey body size relationships. The timing of the sea ice retreat (TSR) also plays a crucial role in the seasonally ice-covered marine ecosystem, because it is tightly coupled with the timing of the spring bloom. Thus, it is important to monitor the temporal and spatial distributions of a phytoplankton community size structure. Prior to this study, an ocean colour algorithm was developed to derive phytoplankton size index F_L, which is defined as the ratio of chlorophyll a (chl a) derived from cells larger than 5 µm to the total chl a, using satellite remote sensing for the Chukchi and Bering shelves. Using this method, we analysed the pixel-by-pixel relationships between FL during the marginal ice zone (MIZ) bloom period and TSR over the period of 1998–2013. The influences of the TSR on the sea surface temperature (SST) and changes in ocean heat content (ΔOHC) during the MIZ bloom period were also investigated. A significant negative relationship between F_L and the TSR was widely found in the shelf region during the MIZ bloom season. However, we found a significant positive (negative) relationship between the SST (ΔOHC) and TSR. Specifically, an earlier sea ice retreat was associated with the dominance of larger phytoplankton during a colder and weakly stratified MIZ bloom season, suggesting that the duration of the nitrate supply, which is important for the growth of large-sized phytoplankton in this region (i.e. diatoms), can change according to the TSR. In addition, under-ice phytoplankton blooms are likely to occur in years with late ice retreat, because sufficient light for phytoplankton growth can pass through the ice and penetrate into the water columns as a result of an increase in solar radiation toward the summer solstice. Moreover, we found that both the length of the ice-free season and the annual median of F_L positively correlated with the annual net primary production (APP). Thus, both the phytoplankton community composition and growing season are important for the APP in the study area. Our findings showed a quantitative relationship between the interannual variability of F_L, the TSR, and the APP, which suggested that satellite remote sensing of the phytoplankton community size structure is suitable to document the impact of a recent rapid sea ice loss on the ecosystem of the study region.