The present study examined (a) processing biases for emotional facial stimuli in a sample of 355 4- to 12-year-old non-clinical children, (b) developmental patterns of such biases, and (c) to what extent biases were related to social anxiety and the temperamental trait of behavioral inhibition in children of various ages. Processing biases were assessed with a dot probe task and a dynamic emotion recognition paradigm (i.e., morph task), whereas children's levels of social anxiety and behavioral inhibition were measured by means of parent-report. Results showed that on the morph task children were generally faster in detecting happy faces compared to angry faces, and this effect was not qualified by age, social anxiety, or behavioral inhibition. Further analyses revealed no significant effect of age on bias scores. However, analyses did reveal two classes in the data with one class mainly consisting of younger children and the other class predominantly composed of older children: younger children were in general slower, less accurate, and displayed more variance in their scores on the processing biases tasks than older children. Results of this study underline the need of the development and use of more age-appropriate, non-reaction time-based tasks for measuring processing bias in younger children.