For individual varieties, tolerance of septoria leaf blotch was quantified by the slope of the relationship between disease and yield. Variation in disease severity and the associated yield responses were provided across two sites and three seasons of field experiments. Slopes were fitted by residual maximum likelihood for two contrasting models: (i) a fixed-effects model, where no prior assumptions were made about the form of the variety slopes; and (ii) a random-effects model, where deviations in individual variety slopes away from the mean variety slope formed a normal random population with unknown variance. The analyses gave broadly similar results, but with some significant differences. The random model was considered more reliable for predicting variety performance. The effects of disease were quantified as symptom area and green canopy duration. Models of the relationship between symptom area and yield were site-specific. When site effects were not taken into account, these models had poor predictive precision. Models based on the canopy green area gave robust predictions of yield and were not site-specific. Differences in disease tolerance were detected in a comparison of 25 commercial winter wheat varieties. Tolerance was not detected directly through symptom measurements, but instead through measurements of canopy green area, which provides a measurement of the effects of disease that accounts for differences in canopy size across sites and seasons. The varieties showing greatest tolerance tended to have lower attainable yield than the intolerant varieties. Presence of the 1BL/1RS chromosome translocation, which has been reported to increase radiation use efficiency, appeared to be associated with intolerance.