Stream rehabilitation and enhancement projects in the Norwalk River (urban-forest watershed) and Merrick Brook (agriculture-forest watershed) were evaluated. Instream structure installation, streambank stabilization and meander re-creation were performed 2–5 years before monitoring. Physical, chemical and biological variables were monitored at control, enhanced (treatment sites originally controls), impaired and rehabilitated (treatment sites originally impaired) sites for three field seasons to evaluate the projects and formulate monitoring strategies. Small improvements in local habitat and macroinvertebrate assemblages were observed at rehabilitated sites on the Norwalk River however control conditions were not attained. Changes to stream health were less evident at the reach scale suggesting that watershed processes that form and maintain habitat were too altered for more widespread recovery. A localized sediment source from a failing streambank was eliminated from Merrick Brook protecting the abundant nearby quality habitat, yet fining occurred at the rehabilitation site due to hydraulic changes leading to localized shifts in macroinvertebrate assemblages. Single-season sampling created a useful snapshot to compare enhanced and rehabilitated sites to control and impaired sites. We recommend a tiered sampling strategy where effectiveness monitoring may include a detailed effort at many sites over a short time (as performed here), a relatively low level of detail (e.g. a rapid assessment) at an intermediate number of sites over a short time, and a detailed long-term monitoring at few sites (e.g. before-after-control-impact, BACI). More research is needed to continue the trend of increased project evaluation to advance the science and application of stream restoration. Copyright © 2010 John Wiley & Sons, Ltd.