Equivalent time sampling is an important technology in the field of high-speed optical waveform testing and quality evaluation. It uses low actual sampling rates for exchanging higher bandwidth and vertical resolution; hence, it is incapable of using filtering, averaging, and other methods for equalization when measuring signals with random and discontinuous characteristics. Therefore, herein, a recursive neural network-based equivalent time sampling signal equalization method is proposed. By training the recursive network model, an equivalent time equalizer is established, and the method is validated by processing equivalent time sampling signals of optical digital communication and light detection and ranging (LiDAR) waveforms. The results show that compared with the input waveform, the eye graph related parameters that characterize the quality of optical communication, i.?e., eye height, eye width, and jitter, exhibit considerable improvements. For linear frequency modulation LiDAR signals, enhancing the waveform amplitude spectrum response solves the problem of equalization processing for equivalent time sampling signals.