Objective: At least 3 randomized, placebo-controlled, double-blind studies have supported the efficacy of computerized attention modification programs (AMPs) in reducing symptoms of anxiety in patients diagnosed with an anxiety disorder. In this study we examined patient characteristics that predicted response to AMP in a large sample of individuals diagnosed with generalized social phobia. Method: The sample comprised 112 individuals seeking treatment for generalized social phobia who completed a randomized clinical trial comparing AMP (n = 55) with a placebo condition (i.e., attention control condition; n = 57). We examined the following domains of baseline predictors of treatment response: (a) demographic characteristics (gender, age, ethnicity, years of education); (b) clinical characteristics (Axis I comorbidity, trait anxiety, depression); and (c) cognitive disturbance factors (attentional bias for social threat, social interpretation bias). Results: Results revealed that ethnicity predicted treatment response across both conditions: Participants who self-identified as non-Caucasian displayed better overall response than did Caucasians. The only prescriptive variable to emerge was attentional bias for social threat at preassessment. Participants in the AMP group who exhibited larger attentional bias scores displayed significantly greater reductions in clinician-rated social anxiety symptoms than did their counterparts in the attention control condition. Conclusions: These results suggest that AMP may be targeted to individuals most likely to benefit from these programs. (Contains 4 tables, 4 footnotes and 1 figure.)