This study addresses an empirical study for extracting a negative brain event-related potential called as mismatch negativity (MMN) from children's EEG recordings through our previously proposed method combining wavelet decomposition and independent component analysis (wICA). MMN here was elicited by the uninterrupted sound under an oddball paradigm including repeated stimuli and deviant stimuli. In theory, larger magnitude of deviance of a deviant stimulus may elicit MMN with larger peak amplitude and shorter latency. We showed that wICA extracted such well-defined MMN from ordinary averages over the first 20 single trials of EEG recordings in the experiment although MMN was not well structured in such ordinary averages and did not possess theoretical properties measured from difference wave. Regarding about 11 minutes cost to collect 700 single trials for one child, with wICA it may just take less than one minute to obtain a well-defined child's MMN elicited by an uninterrupted sound.