Background and objectives: Pretreatment attention bias towards threat stimuli has been shown to predict treatment outcomes following exposure-based treatments. The extent of emotional variability experienced during exposure therapy has also been found to predict better treatment outcomes in anxious adults. The present study examined whether pretreatment attention bias towards threat stimuli and greater emotional variability during exposure activities were associated with stronger treatment outcomes in anxious children receiving group-based cognitive-behavioural therapy (CBT). Methods: Twenty-six anxious children completed a visual probe task with emotional faces, followed by a 10-week CBT program in a group format. Children completed weekly within-session exposure activities during the last 5 weeks of group CBT. Results: Pretreatment attention bias towards threat stimuli, greater emotional variability and within-session habituation during exposure activities were significantly associated with reductions in clinician- and/or parent-rated anxiety symptoms after the 10-week CBT program. Treatment responders had significantly higher peak emotional distress ratings during exposure activities. However, threat attention bias and within-session exposure measures were not significantly related. Conclusions: Pre-existing individual differences in attention bias to threat cues and the degree of emotional reactivity experienced during exposure activities are both important independent predictors of treatment outcomes for anxious children receiving group-based CBT.