AbstractInteroceptive accuracy (IAc), that is, the ability to accurately perceive one's own bodily signals, is widely assumed to be a trait, although experimental manipulations such as stress may affect IAc. We used structural equation modeling to estimate the reliability of IAc, and the proportions of individual differences in IAc, explained by a trait and occasion‐specific effects of situation and person‐situation interactions. We assessed IAc in 59 healthy participants (40 women, MAge = 23.4 years) on three consecutive measurement occasions, approximately 1 week apart, in a rest and poststress condition, using a heartbeat counting and a heartbeat discrimination task. The results showed fair temporal stability (intraclass correlation coefficients ≥ 0.38) and good reliability (Mdn = .63; range .49–.83) for both methods. While around 40% of the variance of a single IAc measurement could be explained by a trait, approximately 27% was accounted for by occasion‐specific effects of situation and person‐situation interaction. These results suggest that IAc measures are relatively consistent and that situations and person‐situation interactions impact IAc as measured at a certain point in time. An aggregation across at least two measurements is recommended when using IAc as a trait variable.