A series of ensemble predictability experiments in a setting of the “perfect” model scenario have been conducted using the global coupled community earth system model (CESM), Version 1.1, to evaluate the seasonal predictability of the northern tropical Atlantic (NTA) pattern in the tropical Atlantic variability (TAV). Our analysis of an 86-year control simulation shows that CESM reproduces the annual cycle and interannual variability realistically. In particular, the NTA patterns, extracted as the first modes of the rotated empirical orthogonal function (REOF) analyses, are consistent between the model and observations. A set of the extreme NTA events is selected from the control simulation based on the peaking values of the time series of the NTA mode. Ensemble predictability experiments are conducted to predict each of these events as the “truth” at seasonal lead times with small perturbations of the atmospheric initial states. The correlation and root mean square error (RMSE) of the ensemble mean, as well as the ensemble spread and reliability, are used to assess the prediction skill quantitatively. It demonstrates that the model can forecast the NTA events skillfully at monthly leads up to 9 months. Composite analysis of the predicted positive and negative events is conducted to explore the physical influences of the regional air–sea interaction and the remote forcing from outside the Atlantic basin, such as the El Nino-Southern Oscillation (ENSO). It is shown that the surface latent heat flux anomalies, generated by surface wind anomalies over the northern tropical Atlantic from boreal fall and winter, force the NTA SST anomalies that peak in spring. Most of these wind anomalies are in turn generated by the remote ENSO forcing. As a result, the NTA pattern can be predicted realistically as long as the ENSO events are predictable.