Whether high-involvement work system (HIWSs) contribute to the performance of firms has been a major issue in debates on workplace reform and economic outcomes. Discussion has been polarized between those identifying strong effects of HIWSs and critics. These debates are placed in the context of an analysis of the social relations of production which draws on ealier studies of industrial relations and productivity. It is argued that (1) the HIWS- erformance connectons are less strong than is often implied, but this does not destroy the underlying theme that how employees are managed affects productivity; (2) HIWSs have several outcomes which need to be considered together, not subsumed under the positive sounding label of 'performance'; (3) HIWSs may be the 1990s equvalent of other mechanisms, and thus be simply the current delivery vehicle rather than being causal in a string sense; and (4) reciprocal causation and reinforcement remain important, that is taht 'performance' sustains an HIWS as much as the reverse.