Healthy adults typically display enhanced processing for self-(relative to other-) relevant and positive (relative to negative) information. However, it is unclear whether these two biases interact to form a selfpositivity bias, whereby self-positive information receives prioritized processing. It is also unclear how a blocked versus mixed referent design impacts reference and valence processing. We addressed these questions using behavioral and electrophysiological indices across two studies using a Self-Referential Encoding Task, followed by surprise recall and recognition tasks. Early (P1) and late (LPP) event-related potentials were time-locked to a series of trait adjectives, encoded relative to oneself or a fictional character, with referent presented in a blocked (Exp. 1) or mixed (Exp. 2) trial design. Regardless of study design, participants recalled and recognized more self-than other-relevant adjectives, and recognized more positive than negative adjectives. Additionally, participants demonstrated larger LPP amplitudes for self-relevant and positive adjectives. The LPP self-relevance effect emerged earlier and persisted longer in the blocked (400-800ms) versus mixed design (600-800ms). The LPP valence effect was not apparent in the blocked design, but appeared late in the mixed design (600-1200ms). Critically, the interaction between self-relevance and valence appeared only behaviorally in the mixed design, suggesting that overall self-relevance and valence independently impact neural socio-cognitive processing.