The most reliable approach for timing fungicides to control Rhizoctonia web blight on container-grown azalea has been a fixed calendar date. The purpose of this study was to model periods when a preventive control could be applied in advance of significant increases in leaf blight intensity (LBI) in response to a daily accumulated risk value indicating conducive conditions of temperature (18 or more hours between 20 and 30°C when maximum temperature is less than 35°C), leaf wetness (16 or more hours), and/or rainfall (greater than 6.7 mm above the maximum daily irrigation). Analysis used weather measurements taken every 30 minutes from 11 site-year datasets from 2006 to 2011, and weekly or biweekly LBI assessments. Six developmental site-year datasets were analyzed using receiver operating characteristic (ROC) curve procedures to evaluate models. Only a single model predicted true positives and negatives significantly greater than a 50:50 probability. Subsequently, five site-year validation datasets were evaluated. Similar results were obtained from both the developmental and validation datasets, which produced ROC areas of 0.7715 (P = 0.0120) and 0.8393 (P = 0.0034), respectively. The practical application of weather-based forecasting for management of web blight in nurseries is discussed.