Finger motion gesture recognition has broad application prospects in remote medical care, intelligent wearable devices, and human-computer interaction. However, many challenges still exist in achieving natural and fluent finger motion gesture recognition. In this paper, we proposed a finger motion gesture recognition system based on a double-clad fiber Bragg grating (FBG) sensor array. Six double-clad FBG sensors with different central wavelengths were deployed at various positions on the subjects forearms. The system can recognize the small deformations of different muscle tissues caused by finger motion in real-time, and it will not affect the natural and fluent motion of the fingers. The system achieved high accuracy recognition of nearly 100% of unknown simple gestures and about 85% of complex gestures based on the double-clad FBG center wavelength shift of six known static gestures. The same experiment was conducted using single-mode FBG. The results showed that the double-clad FBG could effectively recognize simple gestures of single-finger motion and complex gestures of multiple-finger motion. In contrast, the single-mode FBG can only recognize gestures of single-finger motion. The double-clad FBG sensor array finger motion gesture recognition system proposed in this paper not only ensures high-precision deformation measurement to recognize different gestures but also has good stability and anti-interference ability. These results indicate that the research work in this paper has great potential in gesture recognition.