Social bees represent an important group of pollinating insects but can be exposed to potentially harmful pesticides when foraging on treated or contaminated flowering plants. To investigate if such exposure is detrimental to bees, many studies have exclusively fed individuals with pesticide spiked food, informing us about the hazard but not necessarily the risk of exposure. Whilst such studies are important to establish the physiological and behavioural effects on individuals they do not consider the possibility that exposure may change over time. For example, many pesticide assays exclude potential behavioural adaptations, such as a rejection of harmful compounds by choosing to feed on an uncontaminated food source, which would behaviourally lower the risk of exposure. Here we conducted an experiment over 10 days in which bumblebees forage on an array of sucrose feeders containing a range of concentrations (0, 2 & 11 parts per billion) of the neonicotinoid pesticide thiamethoxam. We more closely mimic pesticide exposure in the wild by allowing foraging bees to experience i) a range of pesticide concentrations across a chronic exposure period, ii) repeated interactions with the pesticide, and iii) allowing foraging bees to retain the social cues associated with foraging by using whole colonies. We found that the proportion of visits to pesticide-laced feeders increased over time, resulting in the consumption of more pesticide-laced sucrose. After changing the spatial position of each feeder, foragers continued to preferentially visit the pesticide-laced food, indicating that workers can detect thiamethoxam and alter their behaviour to continue feeding from it. The increasing preference for consuming the neonicotinoid treated food, therefore increases the risk of exposure for the colony during prolonged pesticide exposure. Our results highlight the need to incorporate attractiveness of field relevant concentrations of pesticides to foraging bees (and other insect p