An important question in emotion research is what elicits emotions and causes variations in their quality and intensity. Appraisal theories propose a cognitive process through which stimuli are evaluated on a number of different criteria. The combination of results on these criteria determines which specific emotion is elicited. In the present study, we addressed several questions regarding the mechanisms underlying this process, specifically whether appraisal criteria are processed (a) in a fixed sequence, (b) independently of each other, and (c) by different neural structures or circuits. Two appraisal criteria, stimulus novelty and pleasantness, were manipulated with a 3-stimulus oddball paradigm with affective pictures. Electroencephalographic (EEG) markers for the appraisal processes were distinguished using a spatiotemporal clustering analysis. Consistent with theories that assume a fixed sequence of the appraisal process, the analyses revealed early effects of novelty on global field power and the topographical pattern of EEG activity, followed in time by effects involving pleasantness. Moreover, both measures showed significant interactions of novelty and pleasantness in late processing stages (650 ms-800 ms), indicating that the processing of pleasantness depends on the preceding appraisal of novelty. The results of spatiotemporal clustering suggest that the late processing of highly relevant stimuli is not based on a single mechanism, but consists of the initial activation of distinct neural processes to evaluate novel stimuli, followed by activation of different neural mechanisms for the combined evaluation of both novel and highly valenced (i.e., unpleasant or pleasant) stimuli (a distinction that is not apparent in conventional event-related potential measures).