Abstract Background Considering the high likelihood of chronicity, it is imperative to understand the risk factors and outcomes associated with severe anorexia nervosa (AN), for which Danish registers provide a unique opportunity. We developed a measure of AN severity adapted from clinical literature for use in register-based research. Methods The study population included all Danish individuals born between 1963 and 2007 who were diagnosed with AN from 1969 to 2013. Using register data, we constructed the anorexia nervosa register-based severity index (AN-RSI), incorporating early or late illness onset, number of inpatient admissions and outpatient treatments, cumulative treatment length, and illness duration, each weighted based on clinical importance. Associations between AN-RSI scores, evaluated 5 years after first AN diagnosis, and mortality were estimated using survival analysis. Results Among 9167 individuals diagnosed with AN, 132 died during follow-up: 17 from AN, 30 from suicide, and 85 from other causes. Higher AN-RSI scores were associated with increased rates of mortality from AN, somatic anorexia diagnosis, suicide, alcohol-related causes, and any cause. AN cases who scored in the top 20% of AN-RSI had especially high mortality rates. Furthermore, severe AN cases were also more likely to be in treatment in the next 5 years after severity was established. Conclusions AN-RSI effectively captures mortality and long-term treatment in the absence of detailed patient records and is associated with later mortality in AN patients. AN-RSI could serve as a tool to examine epidemiological and genetic risk factors associated with AN course and outcomes.