AbstractWe describe an interactive procedure, implemented on a PC, for solving multiple‐objective non‐linear programming (MONP) problems. The procedure is composed of two stages. The first stage consists of modelling decision maker's (DM's) preferences using a fuzzy outranking relation in a sample of non‐dominated points. The relation is then translated into a preference cone defining a subset of non‐dominated points of greatest interest to the DM. In the second stage the subset of non‐dominated points generated by the cone is scanned interactively. A graphical interface enables a visual interaction. The DM can either reduce directly the subset of non‐dominated points to be scanned or retract to the first stage and give finer preference information which results in contraction of the cone and reduction of the subset of non‐dominated points scanned in the second stage. Description of the procedure is followed by presentation of its application to a didactic