How can states credibly make and keep agreements when they are uncertain ab
out the distributional implications of their cooperation? They can do so by
incorporating the proper degree of flexibility into their agreements. I de
velop a formal model in which an agreement characterized by uncertainty may
be renegotiated to incorporate new information. The uncertainty is related
to the division of gains under the agreement, with the parties resolving t
his uncertainty over time as they gain experience with the agreement. The g
reater the agreement uncertainty, the more likely states will want to limit
the duration of the agreement and incorporate renegotiation. Working again
st renegotiation is noise-that is, variation in outcomes not resulting from
the agreement. The greater the noise, the more difficult it is to learn ho
w an agreement is actually working; hence, incorporating limited duration a
nd renegotiation provisions becomes less valuable. In a detailed case study
, I demonstrate that the form of uncertainty in my model corresponds to tha
t experienced by the parties to the Nuclear Non-Proliferation Treaty, who a
dopted the solution my model predicts.