The interpretation of postmortem heart weight is often difficult, and references for normal heart weight are important. However, to assess the cause of death at a medicolegal autopsy it is also important to have references based on an unselected population of medicolegal autopsy cases with non-natural causes of death (not due directly to disease). We aimed at studying and deriving references for adult heart weight by considering sex, age and body size in cases with an external cause of death. We identified all medicolegal autopsies in Sweden from 1999 to 2013 (n=79,778) and included 27,645 cases. We applied multivariate piecewise linear regression models in three strata of body mass-underweight, normal-/overweight and obesity. We observed that approximately 50% of the variation in heart weight was explained by age, sex and body size. These variables were slightly less important in explaining the variation in heart weight in the underweight and obese compared to in those normal or overweight. Based on the linear regression models we present equations to calculate the predicted heart weight with reference intervals using age, sex, body weight and height. We provide an online heart weight calculator (http://lundforensicmedicine.com) based on these equations. In the forensic interpretation of postmortem heart weights, we suggest that heart weight references derived in cases with an external cause of death is an important complement to references solely based on healthy and normal hearts. Furthermore, the heart weight references presented are derived from a large population, with sufficient numbers for separate models in underweight, normal-/overweight and obese populations.