In many applications, such as medical disputes at Law, interest lies in whether an observed outcome in an individual case was caused by a specific exposure. An important approach aims to evaluate the probability that the outcome in this case was in fact caused by the exposure, using statistical data on similar individuals. Even the best possible experimental data on exposure and outcome typically can not identify this “probability of causation” (PC) exactly, but can only supply bounds for it. These bounds can be significantly improved by accounting for information about internal processes, using additional variables such as covariates and mediators. This work extends such bounds to the case where we have an instrumental variable.