Purpose: The purpose of this paper is twofold: first, to present patient-level utilisation patterns of hospital-based mental health services funded by private health insurers; and second, to examine the implications of the findings for planning and delivering private mental health services in Australia. Design/methodology/approach: Analysing private health insurance claims data, this study compares differences in demographic and hospital utilisation characteristics of 3,209 patients from 13 private health insurance funds with claims for mental health-related hospitalisations and 233,701 patients with claims for other types of hospitalisations for the period May 2014 to April 2016. Average number of overnight admissions, length of stay and per patient insurer costs are presented for each group, along with overnight admissions vs same-day visits and repeat services within a 28-day period following hospitalisation. Challenges in analysing and interpreting insurance claims data to better understand private mental health service utilisation are discussed. Findings: Patients with claims for mental health-related hospitalisations are more likely to be female (62.0 per cent compared to 55.8 per cent), and are significantly younger than patients with claims for other types of hospitalisations (32.6 per cent of patients aged 55 years and over compared to 57.1 per cent). Patients with claims for mental health-related hospitalisations have significantly higher levels of service utilisation than the group with claims for other types of hospitalisations with a mean length of stay per overnight admission of 15.0 days (SD=14.1), a mean of 1.3 overnight admissions annually (SD=1.2) and mean hospital costs paid by the insurer of $13,192 per patient (SD=13,457) compared to 4.6 days (SD=7.3), 0.8 admissions (SD=0.6) and $2,065 per patient (SD=4,346), respectively, for patients with claims for other types of hospitalisations. More than half of patients with claims for mental health-related hospitalisations only claim for overnight admissions. However, the findings are difficult to interpret due to the limited information collected in insurance claims data. Practical implications: This study shows the challenges of understanding utilisation patterns with one data source. Analysing insurance claims reveals information on mental health-related hospitalisations but information on community-based care is lacking due to the regulated role of the private health insurance sector in Australia. For mental health conditions, and other chronic health conditions, multiple data sources need to be integrated to build a comprehensive picture of health service use as care tends to be provided in multiple settings by different medical and allied health professionals. Originality/value: This study contributes in two areas: patient-level trends in hospital-based mental health service utilisation claimed on private health insurance in Australia have not been previously reported. Additionally, as the amount of data routinely collected in health care settings increases, the study findings demonstrate that it is important to assess the quality of these data sources for understanding service utilisation.