This paper investigates the role of extreme oil price increases in empirical studies of the macroeconomics of oil prices. The innovative approach of rolling impulse responses is applied and data on both the aggregate and the industry-level is considered. The results show that the first oil crisis drives long-run results and superimposes both subsample and industry-specifics. Furthermore, there is evidence that the non-occurrence of large oil shocks after the mid-1980s is an important explanation for the Great Moderation.