We investigated controls on stream sediment denitrification in nine headwater streams in the Kalamazoo River Watershed, Michigan, USA. Factors influencing denitrification were determined by using experimental assays based on the chloramphenicol-amended acetylene inhibition technique. Using a coring technique, we found that sediment denitrification was highest in the top 5 cm of the benthos and was positively related to sediment organic content. To determine the effect of overlying water quality on sediment denitrification, first-order stream sediments were assayed with water from second- and third-order downstream reaches, and often showed higher denitrification rates relative to assays using site-specific water from the first-order stream reach. Denitrification was positively related to nitrate (NO3 ) concentration, suggesting that sediments may have been nutrient-limited. Using stream-incubated inorganic substrata of varying size classes, we found that finer-grained sand showed higher rates of denitrification compared to large pebbles, likely due to increased surface area per volume of substratum. Denitrification was measurable on both inorganic substrata and fine particulate organic matter loosely associated with inorganic particles, and denitrification rates were related to organic content. Using nutrient-amended denitrification assays, we found that sediment denitrification was limited by NO3- or dissolved organic carbon (DOC, as dextrose) variably throughout the year. The frequency and type of limitation differed with land use in the watershed: forested streams were NO3- -limited or colimited by both NO3- and DOC 92% of the time, urban streams were more often NO3- -limited than DOC-limited, whereas agricultural stream sediments were DOC-limited or co-limited but not frequently limited by NO3- alone.