Agricultural lands in colonial New England, U.S.A, were once widespread; 80% of the state of Connecticut (CT) was deforested during the 1800’s. Today, forests have recovered and CT is over 60% forested. These agricultural land use legacies may continue to impact stream ecosystems through structural legacies (e.g. channelization, soil and forest structure) and signal legacies (e.g. nutrients retained in soil and groundwater). Our goal was to evaluate if legacy impacts still exist in watersheds with recovered agricultural land. We sampled 14 small afforested watersheds (< 15 km2) across a historical agriculture land use gradient. We (1) analyzed baseflow surface water and groundwater chemistry for anions, cations, nutrients, alkalinity, conductivity, and pH; (2) collected structural data at the stream reach scale: stream substrate, geomorphic transects, large woody debris, canopy cover, and soil characteristics, and (3) measured stream biointegrity metrics based on macroinvertebrate communities. We found that even in predominantly afforested headwaters, historic (1934) and contemporary (2016) agricultural land use were strongly correlated and thus their relative influence on stream ecosystems was difficult to disentangle. Our initial findings suggest that the 1934 crop and 2016 hay land use land cover (LULC) were the strongest predictors of water chemistry across five baseflow sampling dates. We found that neither contemporary nor historic LULC were strong predictors of stream structural characteristics. The biological condition gradient (based on the macroinvertebrate community) was not correlated with contemporary land use and only weakly correlated with 1934 cropland. Our next steps include incorporating additional watershed characteristics into our analysis (e.g., watershed slope, geologic characteristics) and evaluating the agricultural LULC change patterns (i.e., crop to hay) within sampling watersheds.