Understanding whether high healthcare costs for individuals persist over time is critical for the development of policies that aim to reduce the prevalence of high cost patients. And while high healthcare costs will occur in any given year based on the prevalence of certain morbidities and acute conditions, a large random component of the distribution means that it is rarely the same people driving the bulk of healthcare expenditures. Using administrative data for over 250,000 Australian residents for the years between 2006 and 2011, we analyse the persistence of high annual healthcare costs. We examine the prevalence of high cost persistence in this sample, and then, we use endogenous switching models to identify the morbidity groups that are related with high cost persistence. These models also measure cases of cost amplification that are associated with a history of high cost healthcare. This analysis uses data from multiple categories of healthcare, specifically medical services, pharmaceuticals and admitted patient care. While there is a relatively low number of patients with persistent high cost (approximately 3% of the sample), this group accounted for 19% of aggregate expenditure. Pharmaceuticals were the most persistently high cost category of healthcare with 5% of the sample accounting for 32% of aggregate pharmaceutical expenditure. The morbidities associated with notable cost amplifications are morbidities that are hard to prevent or involve escalations of adverse health states that are difficult to avert. This casts doubt on whether broad policies can reduce the prevalence of individuals with persistently high healthcare costs.