One possible explanation for tropical sea surface temperature (SST) interannual variability is that it can be accurately described by a linear auto-regressive model with damped coupled feedbacks and stochastic forcing. This auto-regressive model can be viewed is a “null hypothesis” for tropical SST variability. This paper advances a new coupled general circulation model (CGCM) coupling strategy, called an interactive ensemble as a method to test this null hypothesis. The design of the interactive ensemble procedure is to reduce the stochastic variability in the air-sea fluxes applied to the ocean component while retaining the deterministic component of the coupled feedbacks. The interactive ensemble procedure uses multiple realizations of the atmospheric GCM coupled to a single realization of the ocean GCM. The ensemble mean of the atmospheric GCM fluxes are applied to the ocean model thereby significantly reducing the variability due to internal atmospheric dynamics in the air-sea fluxes. If the null hypothesis is correct, the SST variability is reduced, and the auto-regressive model defines how much the variability should be reduced. In order to test the null hypothesis, we apply the interactive ensemble procedure to a heuristic coupled model. We then use the heuristic coupled model to interpret the CGCM interactive ensemble results with respect to: (i) SST variance and (ii) how the amplitude of atmospheric internal dynamics depends on the evolving background SST anomaly. There are significant regions where the heuristic model fails to reproduce the CGCM results, suggesting that aspects of tropical Indo-Pacific variability in the CGCM cannot be explained by damped coupled feedbacks and stochastic forcing. These regions are largely coincident with regions of large convective anomalies. Surprisingly, we also find significant regions in the tropical eastern Pacific where the variability due to internal ocean dynamics cannot be neglected.