As a major foodborne pathogen, Campylobacter jejuni receives much attention in quantitative risk assessment. To date, all dose–response assessments have been based on a single human feeding study which unfortunately provides incomplete and possibly biased information on the dose–response relation. An incident at a dairy farm, where several children from a school class became ill as a result of drinking raw milk contaminated with C. jejuni, appeared to show a very clear dose–response relation between the amount of milk consumed and the attack rate. This relation was very nearly exponentially shaped and, therefore, seemed to conflict with the rather slowly rising dose–response relation established in the feeding study. Here we show that both datasets can be reconciled when illness and infection are considered separately. This not only provides new information on the illness dose–response relation for Campylobacter, but also amends the infection dose–response relation because of their conditional dependence.