This paper is the first to apply an econometric shift-share model to tourism. The approach allows us to isolate the growth contributions of changes in regional touristic attractiveness from those induced by the structure of visitors, but does not share the caveats of the conventional shift-share approach. Our application to regional tourism in Austria reveals important results: First, differences in long-run performance between regions are mostly related to idiosyncratic changes in the tourist appeal of individual regions rather than a result of more or less favourable structures of visitors. Second, none of several mega-events during the period observed seem to have left prolonged positive effects on the tourism performance of the host regions. And third, performance appears uncorrelated with tourism intensity of a region. Thus, from a policy and destination management perspective, tourism authorities and local suppliers should mainly focus on upgrading the permanent destination attractiveness rather than investing too much effort into landing mega-events or targeting the visitor mix towards promising source markets.