Cyanobacteria harmful algal blooms (cyanoHABs) are a global concern for aquatic ecosystem and human health. Limited funding for monitoring programs and inconsistent determination of cyanoHAB occurrence present challenges for identifying commonly effective variables for characterizing cyanoHABs and the development of generalized models. We compiled a combination of water quality measurements, lake morphology, climatology, remote sensing data, and observations of cyanoHAB occurrence across New York State and used this dataset to develop two sets of predictive models. The first model predicted chlorophyll a, a common indicator of algal biomass, and assessed the importance of variables for modeled predictions. The most important variables were then used in a second set of models to classify cyanoHAB occurrence. The irradiance attenuation coefficient (K_d), which was estimated from Secchi depth measurements, and total phosphorus were the two most important variables for predicting chlorophyll a. The second model examined several variables for their ability to classify cyanoHAB occurrence. Predicted cyanoHAB occurrence based on thresholds of chlorophyll a, K_d, or total nitrogen all had moderate agreement and were able to correctly classify approximately 70% of observed cyanoHABs. Our analysis indicated that multiple data types were important for predicting chlorophyll a statewide and that simple widely available water quality parameters could classify cyanoHABs occurrence with reasonable accuracy. Identifying variables that can be monitored with increased frequency and decreased latency to detect cyanoHAB occurrence will better inform water managers and provide valuable additional data for further refining predictive models of the likelihood of cyanoHABs occurrence.