High external nutrient loads from agricultural runoff have led to persistent and highly toxic algal blooms in Grand Lake St Marys (GLSM) for decades. These pervasive blooms are concurrent with long-term (2009 – 2021) toxin and environmental monitoring, providing a robust weekly dataset for modeling microcystins. Median weekly microcystin concentrations (23.2 µg/L) routinely exceeded World Health Organization recreational limits (20 µg/L) for the study period (ranged 0.03 - 185.0 µg/L). Here, we used a Bayesian hierarchical dynamic linear model to hindcast weekly microcystin toxins using external nutrient loads from tributary data as well as internal lake nutrient and physicochemical concentrations. Overall, lake TN was the biggest driver of microcystin concentration in GLSM. Likewise, TN:TP was a strong negative driver of microcystin (i.e. low N:P ratios align with lower total microcystins), suggesting that N availability directly impacts toxins. External nutrient loading was positively related to microcystin during winter and spring; however, there was no relationship detected between toxin and external loading during summer or fall (particulate phosphorus exhibited the strongest signal but all external nutrients were unsurprisingly correlated). This lack of direct correlation on a weekly timescale between external loads and cyanobacterial toxins during the summer months likely results from nutrient saturation and reflects the importance of internal loading for bloom maintenance as supported by the correlation between in-lake TN and microcystin. Thus, management goals to reduce the highest biomass and toxins in the summer should focus on reduction of winter and spring external nutrient loads. Supporting this, both 2010 and 2021 had lower rain in the first half of the year (winter/spring), resulting in less loading, and experienced smaller/later low toxicity blooms. This suggests that, although internal nutrient loads are important for bloom maintenance, reduced external loads are an effective management strategy even in nutrient saturated systems such as GLSM.