Vegetated buffers are used extensively to manage wetland-dependent wildlife. Despite widespread application, buffer utility has not been experimentally validated for most species. To address this gap, we conducted a six-year, landscape-scale experiment, testing how buffers of different widths affect the demographic structure of two amphibian species at 11 ephemeral pools in a working forest of the northeastern U.S. We randomly assigned each pool to one of three treatments (i.e., reference, 100m buffer, 30m buffer) and clearcut to create buffers. We captured all spotted salamanders and wood frogs breeding in each pool and examined the impacts of treatment and hydroperiod on breeding-population abundance, sex ratio, and recapture rate. The negative effects of clearcutting tended to increase as forest-buffer width decreased and be strongest for salamanders and when other stressors were present (e.g., at short-hydroperiod pools). Recapture rates were reduced in the 30m, but not 100m, treatment. Throughout the experiment for frogs, and during the first year post-cut for salamanders, the predicted mean proportion of recaptured adults in the 30m treatment was only 62% and 40%, respectively, of that in the reference treatment. Frog sex ratio and abundance did not differ across treatments, but salamander sex ratios were increasingly male-biased in both cut treatments. By the final year, there were on average, only about 40% and 65% as many females predicted in the 100m and 30m treatments, respectively, compared to the first year. Breeding salamanders at short-hydroperiod pools were about 10% as abundant in the 100m versus reference treatment. Our study demonstrates that buffers partially mitigate the impacts of habitat disturbance on wetland-dependent amphibians, but buffer width and hydroperiod critically mediate that process. We provide the first experimental evidence showing that 30-m-wide buffers may be insufficient for maintaining resilient breeding populations of pool-dependent amphibians, at least during the first six years post-disturbance.