Much of the work on polygenic risk score (PRS) analysis focuses on maximizing the prediction of case status at a group level. However, the scenario confronting clinicians is more commonly that an individual presents with symptoms that might fit more than one diagnostic criterion. While PRS summarize genetic risk across the life span, comorbidity and changes to presentation over time are common. Thus, the relationship between PRS results and presenting problem at a given time is not straightforward and approaches to incorporate PRS into clinical practice need to consider differential diagnosis, comorbidity and temporal agnosticism. Using data from the Australian Genetics of Depression and Genetics of Bipolar studies, we consider different ways that PRS could be used to aid diagnosis comparing first-past-the-post, clinical-threshold and profile-based approaches. When comparing PRS for depression and bipolar first-past-the-post approaches ‘correctly’ classified 52.2% of participants with bipolar and 53.9% of those with depression. However, when comparing PRS for depression, bipolar and schizophrenia the proportion of participants ‘correctly’ classified dropped to 38.9% and 42.2% respectively. Taking a clinical-threshold approach using a 5% threshold derived from an unselected Australian population, only 13.2% of those with bipolar met the threshold for the bipolar PRS and 8.7% of those with depression met the threshold for the depression PRS. However, profile-by-group analyses of genetic risk showed distinct profiles for depression and bipolar (P = 3.76x10-17). Similar to many other diagnostic aids in psychology and psychiatry, profile-based approaches show promise for translation of PRS into clinical mental health practice.