Aims: This study investigated the potential of the complexity of ECG time series from healthy participants in terms of the detection of crucial events to classify emotions defined by different levels of valence and arousal. Complexity measures have always been assessed within the research of healthy and pathologic physiological signals, where healthy signals possess a high level of complexity while pathologic ones lose a significant level of their complexity. However, little is known about the relation between complexity measures in physiological signals, namely heart signals, in pathopsychological applications. Methods: ECG recordings were collected from a wearable sensor for 70 participants during one week by the Korean Advanced Institute of Science and Technology of South Korea. Participants rated their daily emotions in terms of valence and arousal scores in a method known as the experience sampling method. 5-minute segments of labeled ECG time series were preprocessed and analyzed using the crucial events detection method known as multiscaled modified diffusion entropy analysis (MSMDEA) which produces complexity measures pertinent to the presence of crucial events, a novel indicator of complexity in the field of physiological signal analysis. T-test was used to determine significant differences between the complexity measures of the different groups of emotion-labeled ECG signals. Results: High and low classes of valence and arousal were distinct in their corresponding measures of complexity with significant differences between the binary classes for each dimension of emotion at p<0.0001 for the binary valence classes and p<0.001 for the binary arousal classification. It was shown that high valence and high arousal states are up to 6% more complex than their low counterparts. Conclusion: Identifying emotions in terms of complexity and crucial events using novel techniques of analysis enables a robust classification and may characterize underlying psychopathology by monitoring long-term recordings of heart signals from wearable sensors.