After early ideas that saw El Niños as isolated events, the advent of coupled models brought the conception of ENSO as a cycle in which each phase led to the next in a self-sustained oscillation. Twenty-two years of observations that represent the El Niño and La Niña peaks (east Pacific SST) and the memory of the system (zonal mean warm water volume) suggest a distinct break in the cycle, in which the coupled system is able to remain in a weak La Niña state for up to two years, so that memory of previous influences would be lost. Similarly, while the amplitude of anomalies persists from the onset of a warm event through its termination, there is no such persistence across the La Niña break. These observations suggest that El Niños are in fact event-like disturbances to a stable basic state, requiring an initiating impulse not contained in the dynamics of the cycle itself.