In recent years, there has been growing interest in examining the stimulus pull effects on respondent narratives to the Thematic Apperception Test (Murray, 1943) using standardized coding methods such as the Social Cognition and Object Relations Scale–Global Rating Method (SCORS‐G; Stein, Hilsenroth, Slavin‐Mulford, & Pinsker, 2011; Westen, 1995). The present study expands on prior work by examining sources of variance in SCORS‐G dimensions and card pull effect patterns in an adult clinical sample characterized by high psychiatric comorbidity and clinical severity. A sample of 158 adult psychiatric patients in long‐term residential treatment provided narratives to 10 TAT cards (five of which have not previously been studied for pull effects). Cards 2 and 7BM pulled for significantly more adaptive narratives (positive pull), whereas Card 13MF pulled for more pathological stories (negative pull). Like prior studies, variance in cognitive dimensions of the SCORS‐G was most explained by person effects, whereas the largest source of variance for all other dimensions was best explained by a combination of the card and the person effects on the narrative. Finally, exploratory analyses of card pull effects within different gender groups were conducted. The implications of these findings for performance‐based future studies and possible clinical applications of card pull findings are discussed.