Background & Aims: Cognitive load theory assumes that the higher the learner's prior knowledge (i.e., the more expert the learner), the lower the intrinsic cognitive load (complexity) experienced for a given problem. While this is the case in many scenarios, there can be cases in which the converse is also true, resulting in more expert learners reporting higher intrinsic cognitive load than novices for the same problem. This can occur in relation to problems involving complex systems (e.g., ecological systems), for which novices' problem representations may underestimate problem complexity and therefore report lower intrinsic load than experts. This finding is borne out in the current paper. Samples, Methods & Results: In Study 1 with 118 participants from the Black Forest area in Germany, participants with higher levels of forestry and ecological expertise evaluated a problem relating to the restructuring of the Black Forest to adapt to climate change as more complex than did novices. In Study 2 (within-subjects design, n = 66 primary-school students), we conceptually replicated this finding in a domain more typical of cognitive load theory studies, mathematics. We found that higher prior knowledge also reduced the underestimation of the complexity of 'tricky', but frequently used, mathematics word problems. Conclusion: Our findings suggest that cognitive load theory's assumptions about intrinsic load and prior knowledge should be refined, as there seems to exist a sub-set of problem-solving tasks for which the traditional relationship between prior knowledge and reported ICL is reversed.