Disciplines across the social and behavioral sciences have documented the critical role of early peer connections in children's development, yet mechanisms facilitating these connections have not been tested in controlled laboratory settings. The present experiment manipulated the context in which unacquainted pairs of children interacted for the first time. We randomly assigned 288 5- to 11-year-old U.S. children (girls: n = 142, boys: n = 145, nonbinary: n = 1; American Indian or Alaska Native: n = 2, Asian or Asian American: n = 24, Black or African American: n = 8, Hispanic or Latino: n = 17, White or Caucasian: n = 200, Multiracial: n = 33, other: n = 2, not provided: n = 2) to interact in pairs with a same- or different-gender partner who matched their age. Randomly assigned peer interactions occurred in one of three conditions: interaction without support (baseline condition) or interaction preceded by adult support that scaffolded children's active (active experimental condition) or passive (passive experimental condition) participation in a self-disclosure activity. Adult coders rated children's connections during the interaction, and children later rated their feelings about their partner. Results revealed that the experimental conditions facilitated children's connections relative to baseline and eliminated gender ingroup favoritism in children's connections; the active experimental condition was particularly effective relative to the passive experimental condition. With its focus on testing mechanisms of connection, this research introduces a new experimental approach that is capable of informing theories and practices surrounding the formation of young children's peer connections. (PsycInfo Database Record (c) 2025 APA, all rights reserved).