In the current study, we test for the specificity of deficits in fear acquisition and extinction for the anxiety disorders spectrum. We compared fear acquisition and fear extinction learning between a group of patients with either an anxiety disorder (n = 93) or depression (n = 103) attending for treatment in our outpatient center and a sample of healthy control participants (n = 60). To assess the specificity of the predictive validity of extinction learning and safety learning for the outcome of exposure-based treatments, patients additionally underwent disorder-specific cognitive behavior therapy (CBT). We found only very little evidence for differences in safety or extinction learning between healthy controls and patients with anxiety-disorders or depression using both a group-based categorical analytic approach, as well as a trans-diagnostic, dimensional analytic approach. On the contrary, for anxiety patients only, more favorable extinction learning and more favorable safety learning was associated with more favorable treatment outcome. In sum, this specific prediction of treatment outcome in anxiety patients confirms and extends current theoretical models of exposure-based treatments for anxiety dis-orders, but does not support the notion of general extinction learning deficits in the anxiety disorders spectrum.