Although cognitive theories of psychopathology suggest that attention bias toward threat plays a role in the etiology and maintenance of anxiety, there is relatively little evidence regarding individual differences in the earliest development of attention bias toward threat. The current study examines attention bias toward threat during its potential first emergence by evaluating the relations between attention bias and known risk factors of anxiety (i.e., temperamental negative affect and maternal anxiety). We measured attention bias to emotional faces in infants (N = 98; 57 male) ages 4 to 24 months during an attention disengagement eye-tracking paradigm. We hypothesized that (a) there would be an attentional bias toward threat in the full sample of infants, replicating previous studies; (b) attentional bias toward threat would be positively related to maternal anxiety; and (c) attention bias toward threat would be positively related to temperamental negative affect. Finally, (d) we explored the potential interaction between temperament and maternal anxiety in predicting attention bias toward threat. We found that attention bias to the affective faces did not change with age, and that bias was not related to temperament. However, attention bias to threat, but not attention bias to happy faces, was positively related to maternal anxiety, such that higher maternal anxiety predicted a larger attention bias for all infants. These findings provide support for attention bias as a putative early mechanism by which early markers of risk are associated with socioemotional development.