This article presents: (1) arguments comparing the use of Harsanyi and
Selten's risk dominance (a measure of the 'risk' associated with choo
sing an action which supports an equilibrium) and payoff dominance as
equilibrium selection criteria; (2) new experimental evidence which su
ggests the existence of a payoff dominated risk dominant equilibrium i
s a necessary (but not sufficient) condition for coordination failure
and the existence of a tradeoff between risk dominance nad payoff domi
nance; and (3) new experimental evidence which links risk dominance an
d payoff dominance to the speed of convergence to equilibrium.