In this Letter, a miniature wearable Raman spectroscopy system is developed. A wearable fiber-optic probe is employed to help the stable and convenient collection of Raman spectra. A nonlinear partial least squares model based on a multivariate dominant factor is employed to predict the glucose level. The mean coefficients of determination are 0.99, 0.893, and 0.844 for the glucose solution, laboratory rats, and human volunteers. The results demonstrate that a miniature wearable Raman spectroscopy system is feasible to achieve the noninvasive detection of human blood glucose and has important clinical application value in disease diagnosis.