Marine harmful algal blooms (HABs) pose a significant economic burden and public health concern for coastal regions. To aid management with HAB monitoring and mitigation, HAB forecasting models are often utilized to predict bloom occurrence, and they typically incorporate HAB species traits. However, HAB species are diverse, each characterized by unique physiologies, and discerning their traits can be labor intensive. To better constrain the traits of North American HAB species (n= 29), we performed an analysis of 595 historical HAB events from coastal North America, totaling >4000 weeks, and correlated each event with sea surface temperature at time of occurrence. We then constructed species distribution models to discern species thermal traits, with unimodal temperature probability curves characterized for 18 HAB species (6 previously uncharacterized). We found HAB events to be correlated with species’ laboratory-determined thermal optima (Topt), the temperature that produces maximal growth. However, HAB events often occurred below the Topt and over a narrower range of temperatures, suggesting analyses of HAB events could better constrain realized temperature thresholds and improve HAB forecasting models. In an analysis of species traits, HAB species also clustered into distinct groups based primarily on their physiological traits as opposed to morphological or behavioral traits, with toxin and temperature traits explaining much of the variation between species. With these data, we documented the fundamental and realized thermal traits of HAB species, increasing our collective ability to predict and monitor future HAB events.