Wavelet transform has recently been developed to the level of sophistication suitable for application to signal processing in magnetospheric research. We explore this new technique in decomposing signals in the time-frequency domain by first conducting continuous wavelet transform on a test signal to show its ability to resolve multiple-frequency components embedded within white noise of half the amplitude as the signal. We then use this tool to examine the large-amplitude magnetic fluctuations observed during a current disruption event. The results show the current disruption to be a multiscale phenomenon, encompassing low as well as high-frequency components. The lowest-frequency component appears to behave quite independently from the higher-frequency components. The analysis shows for the first time that in current disruption the high-frequency components constitute a broadband excitation with a nonstationary nature, i.e., some oscillations appear to cascade from high to low frequency as time progresses. ¿ 1997 American Geophysical Union