Reducing maladaptive cognitions is hypothesized to constitute an active therapeutic process in multidisciplinary pain programs featuring cognitive-behavioral interventions. A cross-lagged panel design was used to determine whether: a) early-treatment cognitive changes predicted late-treatment pain, interference, activity and mood changes, but not vice versa; b) three cognitive factors made unique contributions to outcome; c) substantial cognitive changes preceded substantial improvements in outcome. Sixty-five chronic pain patients, participating in a 4-week multidisciplinary program, completed measures of pain helplessness, catastrophizing, pain-related anxiety (process factors), pain severity, interference, activity level and depression (outcomes) at pre-, mid- and posttreatment. Results showed that early-treatment reductions in pain helplessness predicted late-treatment decreases in pain and interference, but not vice versa, and that early-treatment reductions in catastrophizing and pain-related anxiety predicted late-treatment improvements in pain severity, but not vice versa. Findings suggested that the three process factors predicted improvements mostly in common. However, little evidence was found that large early-treatment reductions in process variables preceded extensive improvements in pain. Findings replicate those of a recent report regarding cross-lagged effects, and offer support that cognitive changes may indeed influence late-treatment changes in outcomes.