Body size is a key feature of any organism, influencing almost every aspect of its life history. Social insects provide an interesting model to study body size, because they often exhibit a high degree of worker size variation within the colony. Size variation is often studied in the lab, sometimes using commercially purchased colonies; therefore, it is important to test if the size variation found in the lab is representative of natural conditions. Furthermore, the distribution of worker sizes within a colony is generally assumed to be adaptive at the colony level. However, such size variation may also result from poor control over brood development (and weak selection on the trait). Using bumble bee (Bombus impatiens), which display a large amount of worker size variation when raised in the lab, we tested whether (1) workers from lab colonies with commercial queens are more variably sized than workers from field colonies with wild queens, and (2) whether workers are more variably sized than concurrently raised males or queens. We found that the size variation was greater in field than lab colonies, and workers varied more than either reproductive caste. Our results indicate that the amount of variation produced is accessible to selection, possibly actively regulated by nurse behavior. These results also suggest that size variation is not a lab artifact and is common across both laboratory and natural environments. Therefore, evolutionary and ecological implications of size variation can be inferred from lab-reared colonies with consideration for even greater size variation.