Despite continuing research efforts, knowledge of the transmission of the highly pathogenic avian influenza (HPAI) virus still has considerable gaps, which complicates epidemic control. The goal of this research was to develop a model to back-calculate the day HPAI virus is introduced into a flock, based on within-flock mortality data. The back-calculation method was based on a stochastic SEIR (susceptible (S) - latently infected (E) - infectious (I) - removed (= dead; R)) epidemic model. The latent and infectious period were assumed to be gamma distributed. Parameter values were based on experimental H7N7 within-flock transmission data. The model was used to estimate the day of virus introduction based on a defined within-flock mortality threshold (detection rule for determining AI). Our results indicate that approximately two weeks can elapse before a noticeable increase in mortality is observed after a single introduction into a flock. For example, it takes twelve (minimum 11 - maximum 15) days before AI is detected if the detection rule is fifty dead chickens on two consecutive days in a 10 000 chicken flock (current Dutch monitoring rule for notification). The results were robust for flock size and detection rule, but sensitive to the length of the latent and infectious periods. Furthermore, assuming multiple introductions on one day will result in a shorter estimated period between infection and detection. The implications of the model outcomes for detecting and tracing outbreaks of H7N7 HPAI virus are discussed.