Forest floor CO2 efflux (FF(cer)) is an important component of global carbon budgets, but the spatial variability of forest floor respiration within a forest type is not well documented. Measurements of FF(cer) were initiated in mid-March of 1991 and continued at biweekly to monthly intervals until mid-November. Observations were made at 45 sites along topographic gradients of the Walker Branch Watershed, Tennessee including northeast and southwest facing slopes, valley-bottoms, and exposed ridge-top locations. The FF(cer) measurements were made with a portable gas-exchange system, and all observations were accompanied by soil temperature and soil water content measurements. As expected, FF(cer) exhibited a distinct seasonal trend following patterns of soil temperature, but soil water content and the volume percent of the soil's coarse fraction were also correlated with observed rates. Over the entire measurement period, FF(cer) ranged from a typical minimum of 0.8 mumol m-2s-1 to an average maximum near 5.7 mumol m-2s-1. No significant differences in FF(cer) were observed among the ridge-top and slope positions, but FF(cer) in the valley-bottom locations was lower on several occasions. An empirical model of FF(cer) based on these observations is suggested for application to whole-stand estimates of forest carbon sequestration.