<p>Fire blight, caused by <i>Erwinia amylovora</i> (EA), is a devastating bacterial disease affecting pome fruit trees with significant economic impacts worldwide. Effective management of the disease is challenging due to the rapid epiphytic growth of EA on stigmas and its subsequent entry into the hypanthium, creating an exceptionally narrow temporal window for blossom blight intervention. Successful management strongly depends on accurate prediction of infection risk and timely applications of preventive measures. This review examines the history of fire blight forecasting, from early empirical and thermal indices to advanced, biologically informed mechanistic models. We focus on two globally used, empirically derived, mechanistic fire blight forecasting models, i.e. MARYBLYT and CougarBlight. Evaluations across North America, Europe, and the Middle East confirm both models capture key blossom blight drivers, yet the accuracy of prediction remains regionally inconsistent. These inconsistencies arise in part from the models’ inability to accurately capture EA growth on apple stigmas under the complex microclimatic conditions of the orchard, where interactions between environment factors, host and pathogen occur. We critically evaluate these models by examining their biological assumptions, design frameworks, and data requirements, while assessing regional variations in their performance. We also highlight key limitations and propose future directions to improve prediction accuracy. This review underscores the critical need for next-generation forecasting models to enhance predictive accuracy and enable more effective disease management strategies.</p>