An affective priming task was used to examine bias in the processing of threat-related material in 25 clinically anxious compared to 25 matched, non-anxious control children and young adolescents. No significant differences were found between anxious and non-anxious children in terms of priming effects. However, age-related differences were found depending upon the valence of the target, independent of anxiety status. Both younger (7 - 10 years) and older (11 - 14 years) children showed faster response times to pleasant targets when they were preceded by a congruent compared to incongruent stimulus, consistent with a traditional priming effect. For threat target stimuli, older children showed no difference in response latency according to the congruency of the prime-target valence. Younger children, in contrast, showed a reverse priming effect for threat target stimuli, with slower response times for threat-congruent trials than for threat targets preceded by a pleasant prime. Possible explanations for developmental differences in the processing of threat-related material are discussed.