Phenytoin sodium (PS) has a tendency to convert to its base form; phenytoin base (PHT) during manufacturing, packaging, shelf life and in-use conditions that can influence its clinical performance. The objective of the present work was to develop a non-destructive, quick and easy analytical method for quantification of PHT in the drug product. A formulation was prepared to contain the excipients of commercial capsule formulation of PS. The formulation containing either 100% PHT or PS was prepared and these formulations were mixed in different proportion to achieve 0–100% PHT matrices. FTIR, NIR and Raman spectra of samples were collected. Data were truncated and mathematically pretreated before development of partial least squares (PLS) and principal component analysis (PCA) regressions model. The models were assessed by slope, intercept, R, R2, root mean square error (RMSE) and standard error (SEP). The models exhibited good linearity over the selected range of PHT in the formulations with low error as indicated by slope that was close to one and small values of intercept, RMSE and SE. The models of NIR based data were more accurate and precise than Raman data based models as indicated by the low values of RMSE and SE. Prediction accuracy of independent samples containing 25% PHT using NIR models were similar to Raman models. On the other hand, the prediction was more precise for the independent sample containing 5% PHT using NIR data based models compared to Raman data based models as indicated by standard deviation. In conclusion, chemometric models based on NIR and Raman spectroscopies provides a fast and easy way to monitor the disproportionation of PS in the drug products.