As part of the NOAA ECOHAB funded Gulf of Maine Toxicity (GOMTOX)1 project, we determined Alexandrium fundyense abundance, paralytic shellfish poisoning (PSP) toxin composition, and concentration in quantitatively-sampled size-fractionated (20–64, 64–100, 100–200, 200–500, and > 500 μm) particulate water samples, and the community composition of potential grazers of A. fundyense in these size fractions, at multiple depths (typically 1, 10, 20 m, and near-bottom) during 10 large-scale sampling cruises during the A. fundyense bloom season (May–August) in the coastal Gulf of Maine and on Georges Bank in 2007, 2008, and 2010. Our findings were as follows: (1) when all sampling stations and all depths were summed by year, the majority (94% ± 4%) of total PSP toxicity was contained in the 20–64 μm size fraction; (2) when further analyzed by depth, the 20–64 μm size fraction was the primary source of toxin for 97% of the stations and depths samples over three years; (3) overall PSP toxin profiles were fairly consistent during the three seasons of sampling with gonyautoxins (1, 2, 3, and 4) dominating (90.7% ± 5.5%), followed by the carbamate toxins saxitoxin (STX) and neosaxitoxin (NEO) (7.7% ± 4.5%), followed by n-sulfocarbamoyl toxins (C1 and 2, GTX5) (1.3% ± 0.6%), followed by all decarbamoyl toxins (dcSTX, dcNEO, dcGTX2&3) (< 1%), although differences were noted between PSP toxin compositions for nearshore coastal Gulf of Maine sampling stations compared to offshore Georges Bank sampling stations for 2 out of 3 years; (4) surface cell counts of A. fundyense were a fairly reliable predictor of the presence of toxins throughout the water column; and (5) nearshore surface cell counts of A. fundyense in the coastal Gulf of Maine were not a reliable predictor of A. fundyense populations offshore on Georges Bank for 2 out of the 3 years sampled.