In visual crowding, target discrimination strongly deteriorates when flanking elements are added. We have recently shown that crowding cannot be explained by simple low-level interactions and that grouping is a key component instead. We presented a vernier flanked by arrays of vertical lines. When the flankers had the same lengths as the vernier, offset discrimination was strongly impaired. When longer flankers were presented, crowding was weaker. We proposed that crowding is strong when the flankers group with the target (equal length flankers). When the target segregates from the flankers, crowding is weaker (long flankers). To understand the neurophysiological mechanisms of grouping in crowding, here, we adapted the above vernier paradigm to a high-density EEG study. The P1 component reflected basic stimulus characteristics (flanker length) but not crowding. Crowding emerged slowly and manifested as a suppression of the N1 component (after 180ms). Using inverse solutions, we found that the N1 suppression was caused by reduced neural activity in high-level visual areas such as the lateral occipital cortex. Our results suggest that crowding occurs when elements are grouped into wholes, a process reflected by the N1 component.