Many utilities use cellulosic adsorption isotherms to calculate the water content of transformer paper insulation from the water activity of the oil. However, a problem is that this method becomes less precise as the paper degrades because the adsorption isotherms change, and thus the water content of the paper derived from the water content of oil is no longer accurate. When this inaccuracy becomes significant, should the water activity, or the water content of paper, be used to estimate life remaining? In this article, a discussion is given on why the rate of degradation of paper is dependent on water activity, and the potential inaccuracies when using only water content of paper.