Manual acupuncture(MA), as a mechanical action, can be equivalent to an external stimulus to the neural system. To explore the effect of MA on brain activities, we design an experiment that acupuncture at Zusanli acupoint with four different frequencies to obtain electroencephalograph (EEG) signals. Neural system is a complex nonlinear dynamics system that possesses strong nonlinear characteristic, so it is necessary for us to introduce nonlinear analysis methods to study EEG signals. Many studies have demonstrated that the complexity of EEG can reflect the states of the brain function, so we adopt two nonlinear time series analysis methods, i.e. Correlation Dimension and Order Patterns Recurrence Determinism, to investigate the effect of MA on the complexity of brain activities. It is found that the complexity of EEG during acupuncture is obviously higher than that before acupuncture, which can prove that MA has effect on the brain.