This paper introduces the idea of an evolutionarily stable strategy di
stribution, which generalizes the idea of an evolutionarily stable str
ategy; roughly speaking, an evolutionarily stable strategy distributio
n is a finite set of symbiotic strategies which is unaffected by low l
evels of mutation. This idea is then applied to the n-person Repeated
Prisoner's Dilemma, of which the usual Repeated Prisoner's Dilemma is
the special case n = 2. Given some standard assumptions on what mutati
ons are possible, it is shown that if the probability of future intera
ctions is sufficiently large, there are no evolutionarily stable strat
egy distributions. (And hence no evolutionarily stable strategies.) An
example is given of an evolutionarily stable strategy distribution in
the case when the set of possible mutant strategies is restricted. (C
) 1997 Academic Press Limited.