Summary: Concurrent engineering has become a popular approach to product development among manufacturing companies. A lively debate has recently arisen upon whether the technique may be conveniently applied under all circumstances, or whether, in some cases, it may be inferior to other approaches, including the traditional sequential one. The paper proposes an analytical model of concurrent engineering and shows that the main, although not the only, determinants in suggesting the optimal allocation of product and process design effort are the ‘evolution speeds’ of product and process parameters. Evolution speed indicates whether the major part of design uncertainty is reduced early or late in the design process and, among other factors, is related to the problem’s innovative content. It is eventually found that concurrent engineering is optimal in the case of routine design and when the process characteristics are not critical. Under other circumstances other three patterns, termed ‘sequential’, ‘process first’ and ‘co-operative R&D’, appear to be superior.