In this study, time series of sea surface temperature, the number of tropical storms and the number of major hurricanes from the North Atlantic Ocean are examined using bivariate autoregressive time series models. Model parameters are used to diagnose the nature of the interaction between storm activity and sea surface temperature. The methodology employed in uses the well-established concept that a process represented by the time series Yt can be considered causal to another process represented by time series Xt, if the knowledge of the past history of Yt improves the prediction of the future values of Xt beyond what can be achieved from the knowledge of the past history of Xt alone. Among other things, results of this analysis should clarify whether or not the relationship between sea surface temperature and the number storm is the same as the relationship between the sea surface temperature and the number of major hurricanes. Recent studies have shown that the number of major hurricanes in the North Atlantic and other ocean basins is increasing and that this increase is related to increases in sea surface temperature. In most ocean basins, the number of tropical storms and the number of cyclone days remain do not show such an increase.