Keynes (1936) first brought up the relationship between consumption and national income, but Kuznets¡¦observation about the U.S. data was not supported by the Keynes consumption function form. So there are many macroeconomic theories trying to explain the phenomenon observed by Kuznets. This paper uses the way developed by Campbell (1987) to test the permanent income hypothesis suggested by Friedman with Taiwan data. In addition, this paper uses the stochastic cointegration developed by Harris, McCabe, and Leybourne (2002) to re-examine the relationship between consum-ption and national income because the traditional non-stochastic cointegration assumes that the error term is linear and homogeneous, which may be too strong to fit the real world. Besides, this paper compares the nonstochastic cointegration with the stochastic cointegration, and the evidence founded is that the permanent income hypothesis is not supported by Taiwan data with these two methods.