The paper reports productivity differentials of 3-8% in favour of profit-sharing firms in the UK engineering industry. The estimates come from equations in which profit-sharing interacts with factor input levels and/or the firms' technological, organisational and labour-force characteristics, and imply more than a simple incentive effect on work effort, or more 'cooperative' behaviour in given circumstances. Model-selection tests reveal that these models dominate those used in previous work, where profit-sharing enters as a disembodied, Nicks-neutral shift in the production surface. A technological labour relations interpretation of the origin of the gains is suggested, which are found to be asymmetrically distributed. The results question policy measures to encourage profit-sharing which do not take account of its significance in the process of organisational design.