An assessment of the importance of input uncertainty in long-term predictions of lake acidification is presented. Mathematical models that simulate the behavior of hydrologic catchments under acid deposition require input data of precipitation, temperature, and deposition chemistry. In long-term studies of lake acidification it is necessary to hypothesize a scenario of future input to the mathematical models. The present study indicates that uncertainty in precipitation amount and acid concentrations can be very important in the characterization of future lake alkalinity levels. It also suggests that the natural day-to-day variability present in precipitation and other weather variables significantly affects long-term predictions of lake alkalinity in watersheds with short response time, even more than weta nd dry weather cycles do. (Copyright (c) 1989 by the American Geophysical Union.)