Abstract: Based on compressed sensing theory, this paper proposed a channel-randomization method without redundant channels for time-interleaved ADC. The proposed method uses random numbers to determine whether the current channel ADC samples the input signal. If not, this channel will be free. When multiple channels free, one of the free channels will be randomly selected to work. Thus, sub-Nyquist sampling with randomization is realized. Furthermore, based on the measurement matrix and orthogonal-matching-pursuit algorithm, the output data is reconstructed and thus full-data rate of time-interleaved ADC is recovered. The proposed channel randomization method based on compressed sensing for time-interleaved ADC is verified by MATLAB. Under the given timing mismatch, the time-interleaved ADC′s SFDR is improved from 53.1 dB to 65.5 dB, which improved 12.4 dB.