The present study presents a method of assessing and fitting a stock-recruitment relationship. The method is based on a combination of an equilibrium production model and a yield per recruit model. It does not use recruitment data which are, in general, highly dispersed, but production data. Thus, the relationship obtained is not of the parametric form of classical stock-recruitment relationship such as those of the Beverton and Holt (1957) and Ricker (1954), however, in some cases, the resulting stock-recruitment curve resembles models currently used in fisheries science. It is interpreted as an expectation of the recruitment conditionally to stock spawning biomass and stock demographic structure. However, for some realistic values of the parameters, the curve thus obtained describes a loop which suggests that two different recruitment levels may be expected from the same stock spawning biomass level. This theoretical property is illustrated with simulations. As an illustrative example, the method is applied to North-West Atlantic gadoid stocks. The significance of the method, its originality, the validity of the underlying assumptions and their conceptual consequences are discussed.