Controls on methane emission from peatlands in northern Minnesota were investigated by correlation to environmental variables and by field manipulation. From September 1988 through September 1990, methane flux measurements were made at weekly to monthly intervals at six sites in the Marcell Experimental Forest, northern Minnesota (two open bog sites, two forested bog sites, a poor fen, and a fen lagg). Flux was related to water table position and peat temperature with simple correlations at individual sites and multiple regression on all sites together. The effect of water table was also investigated experimentally in ''bog corrals'' (open-ended metal enclosures set in the peak) in which water table was artificially raised to the surface in the driest peatland. Temperature largely controlled variation in flux within individual ecosystems at Marcell, but hydrology distinguished between-site variation. Water table position, peat temperature, and degree of peat humification explained 91% of the variance in log CH4 flux, predicted annual methane emission from individual wetlands successfully, and predicted the change in flux due to the water table manipulation. Raising the water table in the bog corrals by an average of 6 cm in autumn 1989 and 10 cm in summer 1990 increased emission by 2.5¿ and 2.2¿, respectively. Just as expanding the scale of investigation from a single habitat in a wetland to several wetlands necessitates incorporation of additional variables to explain flux (water table, peat characteristics), modeling flux from several wetland regions, if possible, will require the addition of climate parameters. ¿ American Geophysical Union 1993