Understanding consistencies and discrepancies in characterizing diversity and quantity of phytoplankton is essential for better modeling ecosystem change. In this study, eukaryotic phytoplankton in the Pearl River Estuary, South China Sea were investigated using nuclear 18S rRNA and plastid 16S or 23S rRNA genes and pigment analysis. It was found that 18S abundance poorly explained the variations in total chlorophyll a (Chl-a). However, the ratios of log-transformed 18S abundance to Chl-a in the major phytoplankton groups were generally environment-dependent, suggesting that the ratio has potential as an indicator of the physiological state of phytoplankton. The richness of 18S-based operational taxonomic units was positively correlated with the richness of 16S-based amplicon sequence variants of the whole phytoplankton community, but insignificant or weak for individual phytoplankton groups. Overall, the 18S-based, rather than the 16S-based, community structure had a greater similarity to pigment-based estimations. Relative to the pigment data, the proportion of haptophytes in the 18S dataset, and diatoms and cryptophytes in the 16S dataset, were underestimated. This study highlights that 18S metabarcoding tends to reflect biomass-based community organization of eukaryotic phytoplankton. Because there were lower copy numbers of plastid 16S than 18S per genome, metabarcoding of 16S probably approximates abundance-based community organization. Changes in biomass organization of the pigment-based community was sensitive to environmental changes. Taken together, it is recommended multiple methodologies be applied to more accurately profile the diversity and community composition of phytoplankton in natural ecosystems.