Engaging and retaining individuals with psychotic disorders in psychosocial treatments is difficult. Early therapeutic alliance, treatment retention, and 12-month outcomes were examined in a subsample of smokers with a psychotic disorder (N = 178) participating in a healthy lifestyles study comparing a telephone versus face-to-face delivered intervention. Therapeutic alliance was assessed using the Agnew Relationship Measure; primary outcomes were treatment retention and changes in symptoms and health behaviors. Contrary to expectations, early alliance did not predict treatment retention. However, elements of both client- and therapist-rated alliance predicted some clinical outcomes (e.g., higher confidence in the therapeutic alliance at session 1 predicted improvements in 12-month depression). Some modest interactions between early alliance and intervention condition were also identified (e.g., clients initially with lower self-perceived initiative, or higher therapist-perceived bonding benefited preferentially from the telephone-delivered intervention), highlighting the need to further examine the interplay between therapeutic alliance and treatment modality.