Background: Loneliness is common in people with psychotic disorders and associated with reduced health and well-being. The relationship between loneliness in psychosis and health service use is unclear. This study examined whether loneliness predicts increased health care utilization in this population, independently of sociodemographics, health and functioning. Methods: We used cross-sectional data from the Second Australian National Survey of Psychosis. Loneliness was assessed using a single-item question, rated on a 4-point scale (not lonely; lonely occasionally; some friends but lonely for company; socially isolated and lonely). Health service use (past 12-months) was measured by the number of general practitioner (GP), emergency department (ED) and outpatient visits, inpatient admissions, and home visits by mental health professionals. Frequent hospital users comprised those in the top 15% of users of at least two services. Results: Negative binomial regression analysis showed that loneliness was associated with an increased number of GP visits, ED visits and inpatient admissions, only. Socially isolated and lonely survey participants were more than twice as likely (OR = 2.6) of being ‘frequent users’ compared to non-lonely responders. Following stringent adjustment for covariates, loneliness remained significantly associated with being a ‘frequent user’ and showed a non-significant trend to an increased number of GP visits and inpatient admissions. Conclusions: Loneliness is a complex social and personal problem for people with psychosis, related to greater use of some health services. Better strategies for identifying and responding to loneliness in this population have the potential to increase well-being and contain health service utilization costs.