This paper studies, both theoretically and experimentally, frame effects in the context of public good provision. We apply Prospect Theory to read the experimental evidence of a Voluntary Contribution Threshold Game which, depending on the sessions, was framed as a situation in which a) subjects had to make a costly contribution to gain a common prize or b) they had to make a costly contribution to avoid to loose it. By contrast with standard expected utility theory, by which frames have no impact on equilibrium behavior, Prospect Theory predicts more contribution in the latter situation. Our experimental evidence backs up Prospect Theory when the contribution threshold is high (i.e. the public good is relatively more difficult to achieve). When contribution threshold is low, expected utility seems more consistent with the experimental evidence.