Three scales of temporal variability were present in methane (CH4) flux data collected during a 2.5 year (mid-1990-1992) study at a small, poor fen in southeastern New Hampshire. (1) There was a strong seasonality to the fluxes (high in summer); monthly average fluxes range from 21.4 mg CH4 m-2 d-1 (February 1992) to 639.0 mg CH4 m-2 d-1 (July 1991). Annual fluxes were 68.8 CH4 m-2 (1991) and 69.8 g CH4 m-2 (1992). (2) There was interannual variability; distribution of flux intensity was very different from 1991 to 1992, particularly the timing and rapidity of the onset of higher fluxes in the spring. (3) There was a high degree of variability in CH4 flux during the warm season; four successive weekly flux rates in July 1991 were 957, 1044, 170, and 491 mg CH4 m-2 d-1. Fluxes were correlated with peat temperature (r2=0.44) but only weakly with depth to water table (r2=0.14 for warm season data). Warm season fluxes appeared to be suppressed by rainstorms. Along with methan flux data we present an analysis of this temporal variability in flux, using a peatland soil climate model developed for this site. The model was driven by daily air temperature, precipitation, and net radiation; it calculated daily soil temperature and moisture profiles, water table location, and ice layer thickness. Temperature profiles were generally in good agreement with field data. Depth to water table simulations were good in 1992, fair in 1990, and poor in the summer of 1991. Using model-simulated peat climate and correlations to methane flux developed from the field data, simulated methane fluxes exhibited the same three modes of temporal variability that were present in the field flux data, though the model underestimated peak fluxes in 1990 and 1991. We conclude that temporal variability in flux is significantly influenced by climate/weather variability at all three scales and that rainfall appears to suppress methane flux for at least several days at this site. ¿ American Geophysical Union 1994