Field trials tested which components of epidemic development of Puccinia striiformis , the cause of yellow rust, were affected by nitrogen (N) fertilizer applied to winter wheat. Both timing and amount of N were varied to affect canopy size and leaf N content, and to provide a supply of mobile N to the pathogen, by causing fresh N uptake after leaf expansion was complete. No N was applied to control plots. A logistic disease-progress function was fitted to disease-severity data, which were assessed in absolute units. Leaf area and specific leaf N (g N per m 2 leaf tissue) were quantified. Large and highly significant effects of N on the upper asymptotes, or 'carrying capacities' ( c ) were found. Effects on rates and points of inflection of the epidemics were not significant. Early N resulted in larger shoot numbers and leaf area, but disease was also more severe, so that by grain filling, the remaining green leaf areas were larger without N than with N. Later N treatments did not increase canopy size, but did increase symptom area compared with the control. These effects differ from the concept that N affects disease as a result of its effect on canopy growth, and therefore canopy microclimate, and suggest instead a substrate effect. Linear regression revealed that 51% of the observed differences in c were explained by variation in specific leaf N, suggesting that growth of the rust fungus may depend directly on particular components of total leaf N.