Recent work has shown that paleoenvironmental genomics, i.e. the application of genomic tools to analyze preserved DNA in sedimentary records, is a promising approach to reconstruct the diversity of past planktonic communities. This provides information about past ecological and environmental changes. A major advantage of this approach is that individual species, including those that did not leave other characteristic markers, can be identified. In this study, we determined which dinoflagellate marker (i.e. 18S rDNA, dinosterol or dinocysts) provided the most detailed information about the late-Holocene succession of dinoflagellates in an Antarctic Fjord (Ellis Fjord, Vestfold Hills). The preserved rDNA revealed two intervals in the 2750-year-old sediment record. The dinoflagellate diversity was the highest until ∼1850 cal yr bp and included phylotypes related to known dinosterol producers. A lower concentration of dinosterol in sediments <1850 cal yr bp coincided with a community shift towards a predominance of the autotrophic sea-ice dinoflagellate Polarella glacialis, which is not a source of dinosterol. Remarkably, cultures of P. glacialis are known to produce other diagnostic sterols, but these were not recovered here. In addition, conspicuous resting cysts of P. glacialis were not preserved in the analyzed sediments. Overall, dinocysts were rare and the paleoenvironmental genomics approach revealed the highest diversity of dinoflagellates in Ellis Fjord, and was the only approach that recorded a shift in dinoflagellate composition at ∼1850 cal yr bp indicative of a colder climate with more extensive ice cover – this timing coincides with a period of changing climate reported for this region.