With the rapid development of digital media, an increasing number of children are exposed to digital media at an earlier age. Utilizing a longitudinal design with three waves of data from 229 first-grade Chinese primary school students, we employed cross-lagged panel models to examine bidirectional relationships between media use patterns (passive television viewing and active video game playing on any screen-based device) and EF development in Chinese children. Media exposure metrics (screen time and content categorization into educational/negative valence) were parent-reported. EF was measured through attentional control (intraindividual reaction time variability) and behavioral inhibition (accuracy scores) derived from a computerized Go/No-Go task. Our findings reveal complex transactional dynamics: (1) Negative content exposure demonstrated bidirectional relationships with attentional control deficits. Children's exposure to negative content in wave 1 predicted reduced attentional control in wave 2, which subsequently predicted increased negative content media exposure in wave 3. (2) Contrastingly, educational video game content enhanced subsequent attentional control. (3) Total video gaming time emerged as a risk factor for attentional control, with exposure in wave 2 predicting attentional decline in wave 3. (4) EF performance inversely predicted media exposure patterns. Poorer attentional control in wave 1 predicted reduced TV viewing time in wave 2), while poorer control in wave 2 predicted increased educational TV exposure. These results suggest that parents may have monitored the media usage of children with poor attention control. Early behavioral inhibition performance reduced subsequent video game time and negative content exposure. These results indicate that better behavioral inhibition would resist the risk of excessive use of video game. These findings carry crucial implications for developing media guidelines and targeted interventions to intervene maladaptive media-behavior cycles in child development.