Studies have shown that pain acceptance is associated with a better pain outcome. The current study explored whether changes in pain acceptance in the very early treatment phase of an interdisciplinary cognitive-behavioral therapy (CBT)-based treatment program for chronic pain predict pain outcomes. A total of 69 patients with chronic, non-malignant pain (at least 6 months) were treated in a day-clinic for four-weeks. Pain acceptance was measured with the Chronic Pain Acceptance Questionnaire (CPAQ), pain outcomes included pain intensity (Numeric Rating Scale, NRS) as well as affective and sensory pain perception (Pain Perception Scale, SES-A and SES-S). Regression analyses controlling for the pre-treatment values of the pain outcomes, age, and gender were performed. Early changes in pain acceptance predicted pain intensity at post-treatment measured with the NRS (B = -0.04 (SE = 0.02); T = -2.28; p = 0.026), affective pain perception at post-treatment assessed with the SES-A (B = -0.26 (SE = 0.10); T = -2.79; p = 0.007), and sensory pain perception at post-treatment measured with the SES-S (B = -0.19 (SE = 0.08); T = -2.44; p = 0.017). Yet, a binary logistic regression analysis revealed that early changes in pain acceptance did not predict clinically relevant pre-post changes in pain intensity (at least 2 points on the NRS). Early changes in pain acceptance were associated with pain outcomes, however, the impact was beneath the threshold defined as clinically relevant.