Objective: Sudden gains in psychotherapy have been found to predict outcome, but the conditions under which this occurs remain understudied. In the present study, we experimentally examined the effects of processing sudden gains on treatment outcome. Method: As part of a large randomized controlled trial of internet-delivered cognitive behavior therapy for social anxiety disorder (n = 182), we experimentally manipulated therapists' responses to sudden gains. Specifically, we randomized individuals who experienced a sudden gain (n = 52) to either receive (n = 26) or not receive (n = 26) processing of the gain. Results: We found that processed sudden gains were significantly less likely to be reversed compared to unprocessed sudden gains. We also found that individuals with processed sudden gains had lower symptom levels at posttreatment/follow-up compared to individuals with unprocessed sudden gains (M-difference = 15.65, SE = 5.87, p = .023, Cohen's d = 0.84; M-difference = 16.68, SE = 6.36, p = .026, Cohen's d = 1.05; respectively). In addition, individuals with unprocessed sudden gains did not have significantly different symptom levels at posttreatment/follow-up compared to individuals who did not experience a sudden gain during treatment. Conclusions: Our findings suggest that in internet-delivered cognitive behavior therapy for social anxiety disorder, the processing of sudden gains (rather than the gains themselves) leads to upward spirals of improvement that affect long-term outcomes. Replication of these findings in additional studies is needed, and, if replicated, such findings could provide the basis for adding processing of sudden gains to existing therapeutic protocols.