We examine voluntary contributions in a two-stage public good experiment with ‘carryover.’ In two treatments, each subject's second-stage endowment is determined by the return from the public good in the first-stage. We manipulate payoffs across treatments such that, relative to our no-carryover baseline, earnings from either Nash Equilibrium (constant NE) play or Pareto Optimal (constant PO) play are held constant. The remaining two treatments maintain a constant endowment in each stage, but vary the marginal per capita return (MPCR high or MCPR low) to contributions in the second-stage. Our results indicate that carryover increases first-stage contributions. Our implementation of carryover enables us to examine the effects of changing endowments and a wide range of MPCR's. Consistent with the literature, we find that MPCR and endowment effects are important determinants of subject contributions to the group account. While stage 1 contributions tend to increase in the presence of carryover, efficiency levels across both stages fall relative to the baseline. Efficiency levels fall because the maximum earnings possible increase with carryover (due to higher endowments or MPCR levels in stage 2).