Parkinsonian tremor is one of the primary symptoms of Parkinson's disease. The current methods of objective assessment are mainly based on amplitude and frequency. In recent years, the ApEn and Cross-ApEn have been introduced as new methods of nonlinear analysis. They are used to assess the regularity and synchrony of time series. In this paper, a set of wireless data acquisition units were developed to collect 3-axis acceleration data. When PD patients group and the control group performed the instructed tasks, the acceleration signals of their thumb and index fingers were recorded at the meantime. We calculated the ApEn and Cross-ApEn for the acceleration signals and compared them respectively. The mean ApEn of Parkinson's patients were significantly lower than healthy subjects, both for resting tremor and postural tremor. In addition, the mean Cross-ApEn between thumb and index signals of PD were also lower than those of healthy people. The results revealed that the ApEn and Cross-ApEn are effective analysis methods to identify Parkinsionian tremor.