This paper studies an explicitly dynamic evolutionary model of Cournot olig
opoly in which the behavior of firms is based on imitation of success and e
xperimentation. It is proved that only (evolutionary) stable sets are selec
ted by the unique invariant measure of the induced Markov chain as the prob
ability of experimentation tends to zero. Evolutionary stability refers to
a finite population set-up here. We show that stable singletons and symmetr
ic Walrasian equilibria coincide in general. Furthermore, Walrasian equilib
ria can be approximated by stable proper intervals. These results generaliz
e and extend the approach of Vega-Redondo [Vega-Redondo, F., 1997. The evol
ution of Walrasian behavior. Econometrica 65, 375-384]. (C) 2000 Elsevier S
cience S.A, All rights reserved.