Introduction: Patterns of affect-biased attention are related to anxiety and anxiety risk. However, little is known regarding how affect-biased attention develops. Recent work suggests relations with both infant temperamental negative affect and maternal anxiety. The current paper examines potential bi-directional relations between infant attention, infant negative affect, and maternal anxiety to better understand a developmental process that may precede the emergence of anxiety. Method: Infant-mother pairs (N = 333) participated in a multi-site, longitudinal study providing eye-tracking and questionnaire data when infants were 4-, 8-, 12-, 18- and 24-months. A random intercepts cross-lag panel model assessed bi-directional relations between infant attention, infant negative affect and maternal anxiety.Results: Within-person deviations in maternal anxiety were prospectively, negatively related to within-person deviations in infant attention to angry face configurations at every assessment and within-person deviations in infant attention to happy face configurations at the final two assessments. Additionally, within-person deviations in infant negative affect were prospectively, positively related to within-person deviations in infant attention to angry face configurations at 12- and 18-months. Consistent bi-directional relations were not found.Conclusion: Our results suggest that infants do not display a stable bias to threat in the first 24 months of life. Rather, individual differences, in this case maternal anxiety and infant negative affect, shape patterns of attention biases over time. The current results provide an initial understanding of bi-directional relations in affect-biased attention development.