Soot production in turbulent flames represents an urgent issue for applied systems and raises a range of fundamental challenges. Most of the literature effort on this question has been carried out on non-premixed turbulent flames, where mixing plays a dominant role. It is however interesting to see if soot production can be investigated in a premixed turbulent flame configuration, thus eliminating the role of mixing in this process. It is shown in the present article that this can be accomplished by making use of rich swirled ethylene/air flames under perfectly-premixed rich conditions established in a confined combustor operating at atmospheric pressure. Quantitative measurements of the soot volume fraction (fv) are carried out by the Laser Induced Incandescence (LII) technique. Although the soot volume fraction levels are one or two orders of magnitude lower than those found in non-premixed flames, it is shown that detection is feasible and that this configuration may be used to analyze effects on soot production of operating parameters such as the equivalence ratio, the power level or the wall temperature. The observed trends are interpreted numerically by considering an idealized model of a rich premixed burner-stabilized stagnation flame. This configuration is here calculated with a detailed kinetic scheme in combination with a soot sectional model. This description is able to account for some of the trends observed experimentally and indicates in particular how an increase in the wall temperature may increase the soot volume fraction as it is effectively observed in the experiment. The model predicts that the soot volume fraction level increases for high equivalence ratios. This finding is in disagreement with the experimental observations on this swirled flame, which exhibits a maximum soot volume fraction for ϕ = 2.1 for all the considered flame powers. This indicates that complex interactions take place between soot, flame, turbulence and thermal environment and that the investigation of these processes will require a comprehensive turbulent simulation. This may be guided by the database developed in the present experimental effort.