OBJECTIVE: To explore therapeutic alliance between physical therapists and patients with knee osteoarthritis during telephone consultations. Specifically, to describe and compare physical therapist and patient ratings, to determine whether alliance changes over time, and to evaluate whether individual characteristics are associated with alliance. METHODS: We performed a secondary analysis of 84 patients in the intervention arm of a randomized controlled trial who completed 5-10 consultations with 1 of 8 physical therapists via telephone over 26 weeks, involving education, advice, and prescription of a strengthening and physical activity program. Therapeutic alliance was measured after the second (week 4) and final consultations (week 26), using the Working Alliance Inventory-Short Form. RESULTS: Patient and physical therapist ratings of the alliance were high. At week 4, patients rated the overall alliance, and all 3 subscales, higher than therapists. At 26 weeks, patients rated the Task subscale higher than therapists. Patient ratings for the Goal subscale decreased over time, while physical therapist ratings for total alliance and the Bond subscale increased. For patients, the variables of living with others, consulting with a therapist with no previous experience delivering care remotely, having more telephone consultations, and having a higher self-efficacy were associated with greater alliance ratings. Therapists were more likely to perceive a stronger alliance if they had less clinical experience and when treating patients who were younger and had higher self-efficacy. CONCLUSION: Physical therapist perceptions of the therapeutic alliance tended to be lower than those of patients early in treatment, but differences were small and of unclear clinical significance. Some subgroups of patients rated the alliance more strongly than others.