Sclerotinia stem rot (SSR) is the most important disease of oilseed Brassica crops in Norway. Fungicide applications should be aligned with the actual need for control, but the SSR prediction models used lack accuracy. We have studied the importance of precipitation, and the role of petal and leaf infection for SSR incidence by using data from Norwegian field and trap plant trials over several years. In the trials, SSR incidence ranged from 0 to 65%. Given an infection threshold of 25% SSR, regression and Receiver Operating Characteristics (ROC) analysis were used to evaluate different precipitation thresholds. The sum of precipitation two weeks before and during flowering appeared to be a poor predictor for SSR infection in our field and trap plant trials (P = 0.24, P = 0.11, respectively). Leaves from three levels (leaf one, three, five), and petals were collected at three to four different times during flowering from nine field sites over two years and tested for SSR infection with real-time PCR. Percentage total leaf and petal infection explained 57 and 45% of variation in SSR incidence, respectively. Examining the different leaves and petals separately, infection of leaf three sampled at full flowering showed the highest explanation of variation in later SSR incidence (R 2 = 65%, P < 0.001). ROC analysis showed that given an infection threshold of 45%, both petal and leaf infection recommended spraying when spraying was actually needed. Combining information on petal and leaf infection during flowering with relevant microclimate factors in the canopy, instead of the sum of precipitation might improve prediction accuracy for SSR.