The inspection and identification of writing ink are important in the field of forensic science. A Fourier transform infrared spectrometer was used to test 20 types of direct liquid ball pen ink samples, and chemometrics was used to rapidly test the direct liquid ball pen ink. The Fourier transform infrared spectrum data were standardized, and the spectrum was pre-processed using three methods: automatic baseline correction, peak area normalization, and Savitzky-Golay 5-point smoothing. The optimal value of classification K was determined using the sum of squares error(SSE). The samples were analyzed using K-means clustering, and the clustering results were explained. The principal component analysis method was used to verify the results of K-means clustering. The group mean equality test was used to investigate the contribution of principal component variables to the Fisher discriminant analysis (FDA) model, and the FDA discriminant model of straight liquid ball pen ink was constructed. The results show that all the ink samples were clustered into three categories using K-means clustering. The principal components analysis–Fisher discriminant analysis (PCA–FDA) model achieved 100% prediction and classification of different categories of straight-liquid ball pen inks with an accuracy of 100% after cross-validation. Infrared spectroscopy combined with the PCA–FDA model can be used for rapid and accurate inspection and identification of direct liquid ball pen inks.