This article provides experimental evidence on the impacts of irreversibility and imperfect monitoring on the efficiency and the equity of a repeated public goods game. We find that irreversibility and imperfect monitoring both lead to inefficient and unequal outcomes through different channels. Irreversibility lowers public goods contribution in earlier periods and makes the initial-period contribution gap between two players long-lasting. Imperfect monitoring hampers conditional cooperation and persistently reduces group contribution. A finite mixture estimation with conditional cooperators provides a coherent account of the treatment effects.