Automated monitoring of feeding behavior may help detect illness among group-housed calves. We examined whether illness in dairy calves fed with an automated milk feeder changed their feeding behavior, and whether these changes were affected by milk rations. Unweaned Holstein calves (64) were housed in groups of 3-16 animals and fed either high (12L/d or ad libitum) or low (4-6L/d) amount of milk or replacer. Thirty-two calves became spontaneously ill from gastro-intestinal and/or respiratory problems before 21d of age, with no differences between feed level (P>0.10). We used PROC MIXED (with a separate model for each feed level) to test differences between 22 calves that succumbed to only one bout of illness and 22 healthy calves of the same feeding level and age. During the 2d prior to illness detection, there were no differences between sick and healthy calves in any measure of feeding behavior (P>0.10). Following clinically identified illness, sick calves fed high levels of milk or replacer showed a decrease in milk or replacer intake (-2.59 ± 0.7 L/d), a reduced frequency of visits to the milk feeder (-2.43 ± 0.3 visits/d), and an increase in the duration of each visit to the milk feeder (1.66 ± 0.5 min/visit), as compared to the matched healthy calves (P<0.05). This was apparent up to 3d after illness detection. Sick calves fed low levels of milk or replacer only showed a decrease in the duration of each visit to the milk feeder (-1.35 ± 0.2 min/visit; P<0.05) compared to healthy calves, with no differences in intake of milk or replacer or frequency of visits to the milk feeder (P>0.10). Illness results in changes in feeding behavior of calves which can be monitored with automated milk feeding equipment but feeding level affects the types of changes in feeding that occur.