**Abstract** *Making a choice is not easy?! Unravelling the task difficulty of comparative judgement to assess student work.* In the last decades, comparative judgement has been introduced in the field of educational assessment to assess complex skills such as writing or creativity. Assessors are presented with two pieces of student work and asked to indicate which piece of work is of higher quality. Research already evidenced that people are good at making relative judgements on two simple stimuli. Student work is, however, of complex nature. It can be characterized as multidimensional, diverse and highly information-loaded. Pairing such complex objects can easily result in a comparison that puts the relative character of comparative judgement to test. What if an assessor is confronted with two texts of similar quality or with two portfolio pages that show creativity in a totally different way? Although such comparisons provide a threat to the use of comparative judgement for the assessment of student work, research into the difficulty of comparative judgement is lacking. This dissertation tries to fill that gap and aims at (1) examining what contributes to the task difficulty of comparative judgement and (2) exploring to what extent this differs depending on the nature of the objects that are compared. To do so, data of four assessments using comparative judgement on either simple objects (collections of dots, spoken words) or pieces of student work (texts, portfolio pages) were gathered. These data were used to examine the role of three pair characteristics (similarity, low alignability and amount of information), task exposure and the assessor in the task difficulty of comparative judgement. Results point to the crucial role of similarity irrespective of the nature of the objects that are compared: the more similar the objects are, the more difficult comparative judgement becomes. Furthermore, task difficulty also differs across assessors during most assessments. Comparative judgement is more difficult for some assessors than for others. Some assessors are also better in dealing with obstacles to the comparison process (e.g. similar pairs) than others. Evidence is, however, inconclusive about several other aspects that relate to the task difficulty of comparative judgement. Examples hereof are the role of task exposure or low alignability of information. Results on these aspects vary across assessments. In contrast with expectations, these differences are in most cases not related to the nature of the objects that were compared.