A closure empirically determining the relationship between convection and the resolved scale fields is required to close convective parameterization in global or regional climate and NWP models. Most of the convective parameterization schemes nowadays use the quasi-equilibrium assumption proposed by Arakawa and Schubert (1974), which assumes that statistically the generation of convective instability by the resolvable scale processes is in quasi-equilibrium with the removal of convective instability by convection. Recently Zhang (2002) found that such a quasi-equilibrium assumption is not valid in the midlatitude continental convection environment. The main reason for it is the large contribution to the net change of convective instability from the thermodynamic changes of the boundary layer air. This study extends the analysis to the tropical data sets. We will systematically examine the similarities and differences between tropical and midlatitude convection in this respect. We will demonstrate that quasi-equilibrium does not work well in the tropical convection environment either. This raises important questions on the accuracy of the current convection parameterization closures and its possible links to known deficiencies in GCM simulations. Implications of this finding on convection parameterization in both GCMs and simple tropical circulation models will be discussed.