Background: Powdery mildew of cluster bean causes considerable yield loss upto 50-55%. Studies carried on cluster bean powdery mildew are very few and as such there is no information related to epidemiology of the disease. Environmental factors decide the epidemic of the disease and are being used to forecast the disease severity. Methods: Five different dates of sowing of cluster bean were evaluated in a randomized complete block design with four replications at Main Agricultural Research Station, College of Agriculture, Dharwad. Observations on the first appearance of the disease and severity were recorded at weekly intervals using disease scoring scale 0-9 and along with pod yield were considered for statistical analysis. The weather data was used to correlate with the disease severity and correlation matrix was worked out. Further, the observations were converted to % disease index to calculate rate of infection and area under disease progress curve (AUDPC) for each date of sowing. Result: Sowing of crop in first fortnight of September recorded the minimum mean disease severity of 19.10% and recorded pod yield of 4.72 t/ha with lower value of area under disease progress curve (1270.50) as against crop sown in first fortnight of November that recorded the 39.39 % of maximum mean disease severity, 3.82 t/ha pod yield and 2583.88 of area under disease progress curve. The mean maximum temperature was positively correlated (0.89) with % disease index whereas, minimum temperature (-0.87), morning relative humidity (-0.81) and evening relative humidity (-0.88) were negatively correlated with % disease index. Disease progression was started at 35 days after sowing and the maximum ‘r’ value of 0.157 was observed during the II fortnight of September followed by an ‘r’ value of 0.142 during the I fortnight of September whereas, a lower ‘r’ value of 0.009 was observed in I fortnight of September sown crop at 63-69 days after sowing.