We survey what is known about correlations in physical systems. Bell f
ound that locally realistic theories are limited in the types of corre
lations which they allow. The more recent results of Greenberger, Horn
e and Zeilinger and of Hardy have shown that local realism imposes eve
n stronger restrictions than those found by Bell. It is less well know
n that quantum mechanics is also limited in the kinds of correlations
it can produce. This was first discovered by Tsirelson. We discuss Tsi
relson's results and related work by Landau. We conclude by presenting
further conditions which quantum mechanical correlations must satisfy
.