Seasonal patterns of marine net primary production (NPP) are crucial for understanding the marine ecosystem and biogeochemical cycles. Uncoupling of seasonal variations between NPP and phytoplankton chlorophyll-a over different areas has attracted much attention. A variety of remote sensing models were developed for estimating NPP over large spatial and temporal scales from space, but there is not yet an inversion strategy that can be widely used in all seas. In this study, after comparisons with previous works, monthly climatological NPP data over 2003-2020 estimated with the Size-fractioned Phytoplankton Pigment Absorption (aph)-based NPP Model (SABPM) were selected to study the seasonal variability of NPP in the South China Sea (SCS). Results show that seasonality of NPP exhibits large spatial differences and considerable interannual variability. Five regions with significant differences in seasonal variability were identified by cluster analysis in the SCS, including coast along the northern SCS(C), coastal waters off Vietnam (V), area off northwest Luzon (L), northwest part of SCS (NW) and Central basin area (CB). NPP in region C and region V show the most obvious seasonal cycle with maximum in summer and low value in winter, which is mainly affected by river discharge and summer upwelling. Region L shows an opposite seasonal variability with the high value occurred in winter time, and this pattern was found to be related with the strong mixing and upwelling in winter. As for the NW region, NPP shows higher value from May to September than other months, in phase with SST, but the amplitude is weaker than that in region C. Shallow nutricline depth (~20 m in summer) as well as the wind-driven mixing were thought to be the main responsible factors. As affected by the nutrient limited condition combined with a deep nutricline depth, NPP in region CB does not exhibit evident variability throughout the year. Only a weak signal was observed in spring in phase with the Photosynthetically active radiation (PAR). Light driven seasonal patterns of NPP are obvious in most areas of SCS except for region L, region V and a patch area in the western SCS. High correlation coefficients for the seasonal variability between NPP and other factors (including the depth of mixed layer, wind stress, SST, etc.) were found in different parts of the SCS. This observation occurred because the nutrient supply was mainly determined by local dynamics, e.g., wind-induced coastal upwelling and offshore transport. The most anomalous change of the seasonal cycle of NPP coincides with the 2015-2016 ENSO and 2020 IOD events, respectively, which shows obvious effect in region V and region L.