This study examined how individual pretreatment symptoms (depression and anxiety) predicted longitudinal, relational processes in couples therapy (the therapeutic alliance and couple satisfaction). This study included 99 heterosexual couples receiving systemic couples therapy. Partners reported their individual pretreatment symptoms of depression and anxiety before intake. Each member reported their therapeutic alliance with the therapist at the end of sessions 2–8, as well as their couple satisfaction before intake and at the end of sessions 4 and 8. A Latent Profile Analysis revealed four profiles characterizing couples on their pretreatment psychological symptoms, namely both higher (where both members of the couple had higher depressive and anxiety symptoms), female higher, male higher, and both lower profiles. Furthermore, longitudinal, dyadic multilevel models revealed that men in male higher, female higher, and both higher profiles had higher initial alliance levels, compared to men in the both lower profile. In contrast, only female partners in the both higher profile had a higher initial therapeutic alliance than those in the both lower profile. Men in the both higher profile had faster reductions in their alliance, whereas other profiles did not predict changes in male or female alliance. Additionally, the male higher and both higher profiles predicted a lower initial level of couple satisfaction in males but not females. The profiles did not predict changes in couple satisfaction, which may be a result of lowered power due to missing data. Findings were discussed in the context of systemic family theory, with clinical implications for distressed couples.