Using examples, we discuss some aspects of surplus production models. Although the precision of some parameter estimators may appear to be good, the strong asymmetry of confidence intervals and the large impact of the choice of a given formula on those estimators go against this feeling. Models are often formulated with inference to the biological background, but giving a biological meaning to (the estimates of) the parameters may be very dangerous. Fishing effort standardization does not lead necessarily to useful results for management. The fleet dynamics, per se, may not allow equilibrium States. In the context, of general frameworks, where the evaluation of the resource is not the unique objective, surplus production models may be however very flexible tools for fishery analysis, with low parameter requirements.