Textile electrode is flexible, folding, washable and biocompatible with skin. With these advantages, the textile electrodes should be an ideal alternative for electromyogram (EMG) recordings in clinical applications. In this study, a textile electrode system was used for EMG signal recordings and its usability and performance in classifying different arm and hand movements were evaluated through real-time pattern recognition control of virtual myoelectric arm. Our results showed that the offline average classification accuracy over seven able-bodied subjects for all ten movements was 91.81%, compared to 96.54% for six basic movements. The real-time performance in operating a virtual arm was quantified with three measures: selection time, completion time, and completion time. Averagely, the seven subjects achieved a motion selection time of approximately 0.36 s, and a completion time of less than 2 s, and a completion rate of around 82%. These pilot results suggest that the textile electrodes could work well in real-time control of multifunctional myoelectric prosthesis.