Club goods (also called toll goods) are collective goods with the possibility of excluding individuals who fail to contribute.The possibility of exclusion from the consumption of the public good has conflicting effects on welfare : on one hand it reduces the incentives to free ride, on the other hand it reduces the number of beneficiaries of the public good. Many clubs require a minimum number of members to be able to provide their activity (e.g., a farmer association, etc.). This step-level component can either be considered as a threshold for the provision of the club good itself, or as a threshold for maintaining some activity within an existing club. While previous experimental research focused on fundraising to provide non-existing public goods, we adopt in this work the second interpretation: the club good already exists and there is a step level for it's maintenance. Aggregate contributions above the threshold constitute an improvement of the club services that benefit only to the club members. The game admits two Nash equilibria: to contribute the provision point and to contribute nothing. The first equilibrium involves a coordination problem. Furthermore, the game involves a social dilemma, since the social optimum is attained if all agents contribute their endowment. Our baseline treatment is a step level public good game with linear payoff above the threshold without money back guarantee. We compare three levels of provision point. Non-contributors are excluded but are informed about the amount of good produced within the club. Our data shows that contributions are significantly higher when exclusion is feasible and when the provision point is low. For the low provision point with exclusion, subjects overcontribute significantly with respect to the threshold and welfare improves. For the high provision point, exclusion lowers contributions and welfare (compared to no-exclusion). Furthermore, we found that exclusion stabilizes contribution over time. The unravelling of contributions in the baseline treatments does not show up in treatments with exclusion.Club goods (also called toll goods) are collective goods with the possibility of excluding individuals who fail to contribute.The possibility of exclusion from the consumption of the public good has conflicting effects on welfare: on one hand it reduces the incentives to free ride, on the other hand it reduces the number of beneficiaries of the public good. Many clubs require a minimum number of members to be able to provide their activity (e.g., a farmer association, etc.). This step-level component can either be considered as a threshold for the provision of the club good itself, or as a threshold for maintaining some activity within an existing club. While previous experimental research focused on fundraising to provide non-existing public goods, we adopt in this work the second interpretation: the club good already exists and there is a step level for it's maintenance. Aggregate contributions above the threshold constitute an improvement of the club services that benefit only to the club members. The game admits two Nash equilibria: to contribute the provision point and to contribute nothing. The first equilibrium involves a coordination problem. Furthermore, the game involves a social dilemma, since the social optimum is attained if all agents contribute their endowment. Our baseline treatment is a step level public good game with linear payoff above the threshold without money back guarantee. We compare three levels of provision point. Non-contributors are excluded but are informed about the amount of good produced within the club. Our data shows that contributions are significantly higher when exclusion is feasible and when the provision point is low. For the low provision point with exclusion, subjects overcontribute significantly with respect to the threshold and welfare improves. For the high provision point, exclusion lowers contributions and welfare (compared to no-exclusion). Furthermore, we found that exclusion stabilizes contribution over time. The unravelling of contributions in the baseline treatments does not show up in treatments with exclusion.