The correct stability criterion of interchange motions in the magnetosphere has been a question of some controversy. Many researchers hold the view first espoused by Gold [1959>. For a low-&bgr; plasma in a dipole magnetic field, Gold argued that the gradient of pV&ggr;, where p is the plasma pressure, V is the unit flux tube volume, and &ggr; is the adiabatic polytropic index, determines the interchange stability. This notion was further developed by Chandrasekhar [1960>. This common belief was questioned by Cheng [1985>, who was subsequently questioned by Rogers and Sonnerup [1986> and Southwood and Kivelson [1987> on the issue. Is Gold's criterion the necessary and sufficient condition for interchange stability? We shall demonstrate that the answer is ''no'' and provide a more accurate criterion for the dipole magnetic field. We shall also discuss the interchange instability in a plasma distribution with a longitudinal gradient. ¿ American Geophysical Union 1996