Estimates of the output gap ought to be a useful guide for macroeconomic policy, both for assessing inflationary pressures and fiscal sustainability, but their reliability has been called into question by the large revisions which they are often subject to, particularly around turning points. Revisions to OECD published estimates of the output gap around the period of the financial crisis have been exceptionally large, with by far the largest contribution to these revisions coming from the labour-efficiency gap. The current paper investigates a modification to the standard OECD production function method for deriving potential output, which involves an additional cyclical adjustment in the derivation of trend labour efficiency. The additional adjustment helps to reduce the occurrence of large end-point revisions and of sign switches between the initial and final estimates of the labour-efficiency gap. The variables which are most often found to be useful in providing this cyclical adjustment of labour efficiency are manufacturing capacity utilisation and the investment share. However, for a few countries additional variables – house prices and credit – have been used to provide the cyclical adjustment, although this raises an issue as to whether the cyclical adjustment should be limited to a core set of variables to ensure the method remains reasonably homogenous across countries. Recent improvements to the specification of the Phillips curve, which imply a tighter fit between the unemployment gap and inflation, should also reduce end-point revisions to the unemployment gap in future.