Agriculture has influenced southern Appalachian streams for centuries, but recent socioeconomic trends in the region have led to extensive reforestation of agricultural land. Stream ecosystem metabolism might recover from agricultural influence as watersheds undergo reforestation, particularly when shade from terrestrial vegetation is restored. We determined historical (1950) and current (1993) forest cover in 2^nd- and 3^rd-order watersheds in 4 counties of the southern Appalachians using a geographic information system. Streams were placed into landuse categories based on % forested land cover in watersheds and riparian zones. Categories included forested (FOR; >98% forested) and 3 levels of agriculture (AG; ranging from 95% forest to <60% forest) with no change in % forest over the past 50 y, and 2 levels of recovery from agriculture (REC) indicated by reforestation after land abandonment. We selected 3 streams from each category and measured gross primary production (GPP) and 24-h respiration (R₂₄) using a 2-station diurnal O₂ change technique and gas releases to determine reaeration rates. We calculated net ecosystem production (NEP) and the ratio of GPP to R₂₄ (P/R) to compare ecosystem energetics among landuse categories. We measured nutrient concentrations, photosynthetically active radiation (PAR), temperature (degree-days), suspended particle concentrations, and benthic algae (chlorophyll a and ash-free dry mass) to determine if these factors were affected by current or historical agriculture and were correlated with metabolism. Concentrations of inorganic nutrients, PAR, degree-days, suspended solids, and benthic algae were significantly higher in AG streams than in FOR streams. Nutrient and suspended solid concentrations also were higher in REC than in FOR streams, but PAR, degree-days, and benthic algae were similar in REC and FOR streams. GPP varied from <0.1 g O₂ m⁻² d⁻¹ in FOR streams to 1.0 g O₂ m⁻² d⁻¹ in AG streams. GPP was similar in REC and FOR streams, suggesting that shading caused by reforestation might reduce GPP to pre-agricultural levels. R₂₄ was 4 to 20× greater than GPP in all stream types, resulting in highly negative NEP. NEP was less negative in AG streams than in FOR and REC streams. Negative NEP and P/R consistently <1 could have been caused by allochthonous organic matter from remnant forested land (up to 75% forested) in agricultural watersheds. GPP and P/R were strongly correlated with PAR, degree-days, and algal biomass, suggesting that reduced light limited primary production in the streams studied. R₂₄ was positively correlated with nutrient concentrations. Shading caused by reforestation appears to be an important mechanism by which stream metabolism recovers from historical agriculture. Our results provide support for stream restoration efforts focused on developing and maintaining streamside forests.