Sea cucumber is the major tonic seafood worldwide, and geographical origin traceability is an important part of its quality and safety control. In this work, a non-destructive method for origin traceability of sea cucumber (Apostichopus japonicus) from northern China Sea and East China Sea using near infrared spectroscopy (NIRS) and multivariate analysis methods was proposed. Total fat contents of 189 fresh sea cucumber samples were determined and partial least squares (PLS) regression was used to establish the quantitative NIRS model. The ordered predictors selection (OPS) algorithm was performed to select feasible wavelength regions for the construction of PLS and identification models. The identification model was developed by the principal component analysis combined with Mahalanobis distance (PCA-MD) and Scaling to the first range algorithms. In the test set of the optimum PLS models, the root mean square errors of prediction (RMSEP) was 0.45, and correlation coefficients (R2) was 0.90. The correct classification rates of 100% were obtained both in identification calibration model and test model. The overall results indicated that NIRS method combined with chemometric analysis was a suitable tool for origin traceability and identification of fresh sea cucumber samples from nine origins in China.