Producers in western Canada can mitigate the risk of Fusarium Head Blight (FHB) infection and damage to their cereal crops by growing resistant varieties and applying fungicides during critical flowering period. However, fungicide application should not be based only on conventional calendar since FHB occurrence and severity are sporadic and primarily influenced by weather conditions. Weather-based decision-making tools can improve FHB management while also providing significant financial and environmental benefits. Several models have been developed worldwide, with some predicting Fusarium damaged kernels (FDK) or deoxynivalenol (DON) indirectly based on visual estimates of FHB incidence/severity/index (FHBi). This study analyzed FHB over two growing seasons and revealed no significant correlation between FHBi and FDK and FHBi and DON in all crop types except durum. However, the correlation between FDK and DON was significant across all crop types; though, it varied between the two years. Weather-based risk models were developed for predicting FHBi, FDK, and DON in spring wheat, winter wheat, barley, and durum across three Canadian prairie provinces. The number of models developed ranged from 5 to 9 for each disease indicator and crop type, but only two best models for each disease indicator and crop type were further evaluated. The prediction accuracy of the selected models ranged between 75 and 81, 77 and 84, 78 and 79% for FHBi, FDK, and DON, respectively, across crop types. The selected models were validated using producer field data collected in western Canada. The prediction accuracy of the models across crop types ranged between 70 and 100, 66 and 89, 75 and 82% for FHBi, FDK, and DON, respectively. The accuracy of models was greatest when the distance between the fields and nearest weather stations was <40 km. Additionally, this study validated FHBi models currently used in western Canada, which were originally developed in the USA. Although the De Wolf I model predicted winter wheat FHBi with high accuracy (80%), it predicted spring wheat with low accuracy (59%). The models will be used to power an interactive, online digital viewer and provide early warning of potential FHBi, FDK, and DON epidemics in prairie cereal crops.