Dinoflagellate-derived polyketides are typically large molecules (>1000 Da) with complex structures, potent bioactivities, and high toxicities. Their discovery suffers three major bottlenecks: insufficient bioavailability, low-yield cultivation of producer organisms, and production of multiple highly related analogues by a single strain. Consequently, the biotechnological production of therapeutics or toxicological standards of dinoflagellate-derived polyketides is also hampered. Strategies based on sensitive and selective techniques for chemical prospection of dinoflagellate extracts could aid in overcoming these limitations, as it allows selecting the most interesting candidates for discovery and exploitation programs according to the biosynthetic potential. In this work, we assess the combination of data-dependent liquid chromatography coupled with high-resolution tandem mass spectrometry (LC-HRMS2) and molecular networking to screen polyol polyketides. To demonstrate the power of this approach, we selected dinoflagellate Amphidinium carterae since it is commonly used as a biotechnological model and produces amphidinols, a family of polyol-polyene compounds with antifungal and antimycoplasmal activity. First, we screened families of compounds with multiple hydroxyl groups by examining MS2 profiles that contain sequential neutral losses of water. Then, we clustered MS2 spectra by molecular networking to facilitate the dereplication and discovery of amphidinols. Finally, we used the MS2 fragmentation behavior of well-characterized luteophanol D as a model to propose a structural hypothesis of nine novel amphidinols. We envision that this strategy is a valuable approach to rapidly monitoring toxin production of known and unknown polyol polyketides in dinoflagellates, even in small culture volumes, and distinguishing strains according to their toxin profiles.