In this study, by coupling long-term ecological data with empirical proxies of societal demand for benefits, we measured the capacity of forest watersheds to provide ecosystem services over variable time periods, to different beneficiaries, and in response to discrete perturbations and drivers of change. We revisited one of the earliest ecosystem experiments in North America: the 1963 de-vegetation of a forested catchment at Hubbard Brook Experimental Forest in New Hampshire, USA. Potential benefits of the regulation of water flow, water quality, greenhouse gases, and forest growth were compared between experimental (WS 2) and reference (WS 6) watersheds over a 30-year period. Both watersheds exhibited similarly high capacity for flow regulation, in part because functional loads remained low (i.e., few major storm events) during the de-vegetation period. Drought mitigation capacity, or the maintenance of flows sufficient to satisfy municipal water consumption, was higher in WS 2 due to reduced evapotranspiration associated with loss of plant cover. We also assessed watershed capacity to regulate flows to satisfy different beneficiaries, including hypothetical flood averse and drought averse types. Capacity to regulate water quality was severely degraded during de-vegetation, as nitrate concentrations exceeded drinking water standards on 40% of measurement days. Once forest regeneration began, WS 2 rapidly recovered the capacity to provide safe drinking water, and subsequently mitigated the eutrophication potential of rainwater at a marginally higher level than WS 6. We estimated this additional pollution removal benefit would have to accrue for approximately 65–70 years to offset the net eutrophication cost incurred during forest removal. Overall, our results affirmed the critical role of forest vegetation in water regulation, but also indicated trade-offs associated with forest removal and recovery that partially depend on larger-scale exogenous changes in climate forcing and pollution inputs. We also provide a starting point for integrating long-term ecological research and modeling data into ecosystem services science.