In this article, we draw together research on examiner training and on the nature of the judgements entailed in the marking process. We report new analyses of data from two recent empirical studies, Greatorex and Bell (2008) and Suto and Nadas (2008a), exploring possible relationships between the efficacy of training and the complexity of the cognitive marking strategies apparently needed to mark the examination questions under consideration. In the first study reported in this article, we considered the benefits of three different training procedures for experienced examiners marking AS-level biology questions. In the second study reported here, we explored the effects of a single training procedure on experienced and inexperienced (graduate) examiners marking GCSE mathematics and physics questions. In both studies, it was found that: (i) marking accuracy was better after training than beforehand; and (ii) the effect of training on change in marking accuracy varied across all individual questions. Our hypothesis that training would be more beneficial for apparently more complex cognitive marking strategy questions than for apparently simple cognitive marking strategy questions was upheld for both subjects in Study 2, but not in Study 1.