Objectives Medication non-adherence is a major contributor to poor outcomes following discharge from psychiatric hospitals. It is therefore imperative that staff are able to accurately assess the extent to which patients have engaged with treatment. This study presents data on a new observer rating instrument of patient medication adherence. Methods Staff participating in a medication adherence training programme ('Medication Alliance') were given a brief overview of the Observer Rating of Medication Taking (ORMT) scale. Participants then watched six video vignettes of patient adherence behaviour and provided a rating on their scale for each vignette. Participant ratings were then compared with 'expert' ratings. Results Percentage agreement between 'experts' and participants ranged from a low of 68% through to 98% agreement. Only one vignette was rated significantly differently [Mdiff = 0.33, t(49) = 2.08, P = 0.007], with 32% of people falling outside the expert rating range (all of those ratings being higher). This difference was attributed to between group differences, [F(2,47) = 3.49, P < 0.05] and post hoc assessment suggested that the differences between expert and trainee ratings for the vignette may be explained by trainee characteristics, as distinct from inherent characteristics of the rating scale. Conclusion The ORMT can help mental health professionals identify particular non-adherent behaviours thus facilitating identification and treatment of likely non-adherence before discharge. The scale appears to be accessible to a variety of professions with a range of experience and requires minimum training in order to be used reliably.