Self-referential information is uniquely salient and preferentially processed even in children. The literature has used the Self-Referent Encoding Task (SRET) combined with ERPs to study the neural substrates of self-referential processing and its role in development. However, no work has implemented a data-driven, comprehensive examination of the ERP correlates of SRET in youths by comparing a self-referential condition with an other-referential condition. Ninety-two 10-to-14-year-old typically developing youths completed an ERP version of the SRET consisting of a self-referential and an other-referential condition, following which they were unexpectedly asked to complete a recognition task of the presented words. A data-driven Principal Components Analysis isolated five SRET-elicited ERPs: P1, P2, N400, and anterior and posterior late positive potential (aLPP, pLPP). Two-way ANOVAs (Referent × Valence) demonstrated a “self-positivity” bias in aLPP, recognition, and memory sensitivity: youths showed an enhanced aLPP, better recognition, and higher memory sensitivity for Self-Positive versus Self-Negative words, whereas no such differences were found between Other-Positive and Other-negative words. Further, a (marginal) “other-negativity” bias was found in pLPP, P2, and recognition: youths displayed an enhanced pLPP and P2 and higher memory sensitivity in the Other-Negative versus Other-Positive condition, whereas no such pattern was observed in the Self conditions. We provided novel evidence on a self-positivity bias that uniquely favored positive self-referential words as well as an other-negativity bias that uniquely favored negative other-referential words. These findings contribute to our mechanistic knowledge of self-referential processing in youths and inform future studies on the role of self-referential processing in socioemotional development.