Tax Burden, defined as the ratio of total tax revenues over personal income, is frequently used to measure state tax policy. We analyze the empirical relationship between changes in Tax Burden and changes in tax policies from 1987 to 2000 using states’ forecasts of revenue impacts of new tax legislation. Our two major findings have important implications. First, we demonstrate that income-induced, non-tax policy changes are a significant determinant of changes in Tax Burden. These income effects are likely to cause misinterpretation when Tax Burden is used as a variable in economic growth regressions. Second, we estimate that approximately half of the total variation in Tax Burden is due to changes in non-tax policy factors. This finding quantifies the extent of the “mismeasurement” problem which has been discussed, but not analyzed, in previous literature. In concluding, we promote the use of alternative approaches for estimating the economic effects of taxes.