Chemical mass balances for sodium, magnesium, chloride, dissolved organic carbon, and oxygen 18 were used to estimate groundwater seepage to and from Williams Lake, Minnesota, over a 15-month period, from April 1991 through June 1992. Groundwater seepage to the lake and seepage from the lake to groundwater were determined independently using a flow net approach using data from water table wells installed as part of the study. Hydrogeological analysis indicated groundwater seepage to the lake accounted for 74% of annual water input to the lake; the remainder came from atmospheric precipitation, as determined from a gage in the watershed and from nearby National Weather Service gages. Seepage from the lake accounted for 69% of annual water losses from the lake; the remainder was removed by evaporation, as determined by the energy budget method. Calculated annual water loss exceeded calculated annual water gain, and this imbalance was double the value of the independently measured decrease in lake volume. Seepage to the lake determined from oxygen 18 was larger (79% of annual water input) than that determined from the flow net approach and made the difference between calculated annual water gain and loss consistent with the independently measured decrease in lake volume. Although the net difference between volume of seepage to the lake and volume of seepage from the lake was 1% of average lake volume, movement of water into and out of the lake by seepage represented an annual exchange of groundwater with the lake equal to 26--27% of lake volume. Estimates of seepage to the lake from sodium, magnesium, chloride, and dissolved organic carbon did not agree with the values determined from flow net approach or oxygen 18. These results indicated the importance of using a combination of hydrogeological and chemical approaches to define volume of seepage to and from Williams Lake and identify uncertainties in chemical fluxes. ¿ 1997 American Geophysical Union