Despite a recently flourishing literature on the issue, the impact of terror attacks on voters’ behavior is still unclear. In order to identify the causal effect of terrorism, many recent studies have taken an Unexpected Events during Surveys Design (UESD) approach. These studies exploit the occurrence of terror attacks during the fieldwork of surveys, which quasi-experimentally split the sample between respondents interviewed before and after the attack, and thus exposed and not exposed to the terrorist violence. Results of these studies, however, have been mixed. We claim that the reason is that the effect of terrorist attacks is contingent on the specific characteristics of the event and the context surrounding it. In an effort to average out the specificities of individual attacks, this paper leverages the accidental coincidence of over 70 terror attacks perpetrated by ETA during the fieldwork of surveys administered by the main Spanish public opinion institute, involving more than 200,000 interviews. Using this multiple UESD approach, we are able to identify that the causal effect of terrorism predictably varies with certain characteristics of the attack, such as closeness to elections and selectivity of the targets. Our results shed light on the controversy over the electoral effect of terrorism, but are also relevant in pointing out the limits of the identification strategy based on the occurrence of a single, or a small number of unexpected events.