Published data from long-term trials at Moree, New South Wales (1986–1996), and Billa Billa, Queensland (1986–1993), were analysed to determine the factors that influence the incidence of crown rot, caused by Fusarium pseudograminearum, in successive stubble-retained, no-till wheat crops and to examine the feasibility of developing a forecasting system for the disease. Polyetic progress of the epidemics could be described by a form of the logistic growth model with a carrying capacity (K) about 5% higher than the maximum recorded incidence at each site. Infection rate between seasons was positively correlated with yield and in-crop rainfall in the previous season, both of which were indicators of biomass. Infection rate was negatively correlated with rainfall parameters during the summer fallows, which were indicators of conditions favouring residue decomposition. In-crop rainfall, stored soil moisture and temperature parameters were not significantly correlated with infection rates. Multiple regressions based on incidence in the previous season, summer rainfall and either yield or in-crop rainfall in the previous season accounted for 65–81% of the variation in disease incidence at Moree and 86% of the variation in incidence at Billa Billa. Simplified parameters for use in on-farm forecasting systems were explored. The most useful of these was the square root of the product of incidence and either yield or in-crop rainfall, which gave sufficiently accurate predictions at each site to estimate the qualitative risk of crown rot in the following crop. This could be used to decide whether management options such as resistant varieties, rotations or burning were required.