We spend a significant part of our lives navigating emotionally charged mind-wandering states by spontaneously imagining the past or the future, which predicts general well-being. We investigated brain self-generated affective states using EEG microstate analysis to identify the temporal dynamics of underlying brain networks that sustain endogenous affective state activity. With this aim, we compared the temporal dynamics of five distinct microstates between baseline resting-state, positive (e.g., awe, contentment), and negative (e.g., anger, fear) affective self-generated states. We found affect-related modulations of B, C, and D dynamics. Microstates B and D were increased, while microstate C was decreased during negative and positive valence self-generated affective states. In addition, we found valence-specific mechanisms of spontaneous affective regulation. Negative valence self-generated affective states specifically modulate the increased presence of D microstates and decreased occurrence of E microstates compared to baseline and positive valence affective states. The self-generated positive valence affective states are characterized by more prevalent B and les present A microstates compared to both baseline and negative valence affective states. These findings provide valuable insights into the neurodynamic patterns of affective regulation and implications for developing biomarkers for therapeutic interventions in mood and anxiety disorders.