This study consisted of secondary analyses of data from 2 randomized clinical trials to test whether pretherapy cognitions predict CBT outcomes. The sample consisted of 155 primary insomnia patients with sleep maintenance complaints. Of these, 98 were randomized to CBT, 23 were assigned to progressive muscle relaxation training (PMR), and 34 were assigned to a control (sham therapy or wait-list) condition (CON). All patients completed the Dysfunctional Beliefs and Attitudes about Sleep Scale (DBAS), a sleep-related Self-Efficacy Scale (SES) and nightly sleep diaries for 2 weeks prior to receiving their assigned treatment. They then completed sleep diaries throughout an 8-week acute treatment period and during a 2-week period at a posttherapy follow-up. A subset of the sample (n=67) also completed polysomnography immediately before and after completing their assigned treatment. Preliminary regression analyses conducted with a small subset (n=15) of the patients receiving CBT showed those with relatively high levels of unhelpful sleep-related beliefs (Type 1 patients), as reflected by their pretherapy responses to the DBAS and SES questionnaires, showed markedly greater reductions in nocturnal wakefulness in response to CBT than did those (Type 2 patients) reporting less pronounced sleep-related beliefs. Given these findings, we used the regression equation derived from our initial analyses to dichotomize our entire sample into Type 1 (n=82; 52.9%) and Type 2 (n=73; 47.1%) subgroups. Subsequent comparisons showed CBT-treated Type 1 patients had significantly less wake time after sleep onset during most of the 8-week treatment phase than did the Type 1 and 2 individuals assigned to either PMR or CON. Relative to patients assigned to the PMR and CON conditions, CBT-treated Type 1 patients showed better performance across multiple subjective and objective benchmarks of clinically significant improvement, whereas the CBT-treated Type 2 patients did not. Results suggest that insomnia patients' pretherapy cognitive dispositions predict CBT outcome, and those with a pronounced sense of sleep-related helplessness are best suited for this treatment which targets this cognitive stance.