Objective Blood biomarkers of neuronal injury such as neurofilament light (NfL) are being intensively studied to improve diagnosis and treatment of neurodegenerative disorders, but gaps remain in its ability to assist in distinguishing neurodegenerative from primary psychiatric disorders (PPD) with overlapping clinical presentations that commonly cause diagnostic dilemmas. This study aimed to investigate plasma NfL in a range of PPDs, and the diagnostic utility of plasma NfL in differentiating PPD from behavioural variant frontotemporal dementia (bvFTD), a neurodegenerative disorder commonly misdiagnosed initially as PPD. Furthermore, improved understanding of NfL in a diverse range of PPDs, the role and performance of a large normative/reference data sets and models to facilitate precision interpretation of an individual levels, and the influence of covariates, will be critical for future research and clinical translation. Methods Plasma NfL was analysed using Single molecule array (Simoa) technology in major depressive disorder (MDD, n=42), bipolar affective disorder (BPAD, n=121), treatment-resistant schizophrenia (TRS, n=82), and bvFTD (n=22). Comparisons were made between the four clinical cohort groups, and the reference cohort (Control Group 2, n=1926, using generalised additive models for location, scale, and shape (GAMLSS), and age-matched controls (Control Group 1, n=96, using general linear models), Results Large differences were seen between bvFTD (mean NfL 34.9pg/mL) and all PPDs and controls (all <11pg/mL). Plasma NfL distinguished bvFTD from PPD with high accuracy; a 13.3pg/mLcut-off resulted in 86% sensitivity, 88% specificity. GAMLSS models using the large Control Group 2 performed equally to or outperformed models using local controls. An internet-based application was developed to provide individualised z-scores and percentiles based on this reference cohort, which can facilitate precision interpretation of an individual level. Slightly higher plasma NfL levels were found in BPAD, compared to both control groups, and compared to TRS. Conclusions This study adds further evident on the strong diagnostic utility of NfL to distinguish bvFTD from clinically relevant PPDs, and includes the largest cohort of BPAD to date. The finding of higher plasma NfL levels in the largest cohort of BPAD to date should prompt further investigation. Use of large reference cohorts and GAMLSS modelling may have important implications for future research and clinical translation. Studies are underway investigating clinical and diagnostic utility of plasma NfL and the serviceability of the internet-based application for diverse neurodegenerative and primary psychiatric conditions in real-world primary care and specialist clinical settings.