Hysteresis is receiving renewed interest and empirical support but the mechanisms behind it need better understanding. As a novel contribution to such understanding we empirically assess the interpretation according to which long-term unemployment causes hysteresis by increasing the NAIRU. We analyse a panel of 25 OECD countries over a long time span along two lines of enquiry. First, we verify whether long-term unemployment is ‘reversible’. Second, we focus on episodes of strong long-term unemployment reduction to check for inflationary effects in the subsequent years. We find a strong association between total and long-term unemployment rates even in upswings, testifying to macroeconomic reversibility. We do not find accelerating or persi...