Abstract Background: Autobiographical memory style (AM) can serve as a marker for determining susceptibility for depression. Currently and formerly depressed persons show reduced specific memory for autobiographical events. AM also has proven to be a reliable predictor for the remissive course of depression. Recently the issue arose as to whether this predictive power depends on particular dimensions of depression (cognitive, somatic). Methods: To replicate the predictability in vulnerable patients, we tested two groups of obstructive sleep apnea patients (OSAS) before and 2 months after initiation of nasal continuous positive airway pressure therapy (nCPAP). One group had a history of major depression (vulnerable persons; VUL), and the other group did not (NON-VUL). Sensitivity of AM on the dimensions of depression was assessed through two published versions of the cognitive and somatic subscales of the Beck Depression Inventory (BDI). Results: There was a significant interaction of Group by AM recall after presentation of positively valenced cues (AM pos) on the cognitive but not on the somatic dimension of the BDI. Only in the VUL did AM pos predict recovery. Limitations: Generalizability to other patient populations awaits further research. Reliability of the vulnerability assessment prior to the OSAS onset was not established. Conclusions: First, our findings further support the trait assumption of AM. Second, the AM methodology suggests sensitivity to only the cognitive dimensions of depression.