Aim: The psychological construct of 'therapeutic alliance' can be used to better understand the effectiveness of consultations, particularly goal setting for weight management. We analysed audio-recorded health coaching sessions during a weight loss trial to explore relationships between therapeutic alliance and various contextual factors. Methods: Audio recordings of 50 health coaching sessions were analysed. After assessing fidelity to the protocol, therapeutic alliance was measured using an adapted Working Alliance Inventory Observer-rated Short Version (WAI-O-S), and examined by (i) identifying relationships between contextual factors and WAI-O-S scores (Spearman's coefficients); (ii) testing the impact of preparatory exercises and body mass index on WAI-O-S scores (one-way analysis of variance and least-squared differences tests) and (iii) comparing differences in WAI-O-S scores based on relationship status, gender and follow-up session completion (independent samples t-tests). Results: Fidelity was high (mean 88%). WAI-O-S total scores ranged from 55 to 70 (out of 84). Session duration was significantly correlated with WAI-O-S component of 'Bond' (r=0.42, P=0.002). Those who completed preparatory exercises had significantly higher total WAI-O-S scores, 'Goal' and 'Task' scores. Participants who completed the follow-up session scored significantly higher for 'Goal' compared to no follow-up. Conclusions: Spending more time in a session appears related to increased bonding, a key component of therapeutic alliance. Preparatory work may help build therapeutic alliance and agreement on goals appears to influence follow-up completion. These exploratory findings provide directions for research addressing the professional relationship in dietetic consultations for weight loss.