Methodology was developed and evaluated for the rapid detection of castor bean meal (CBM) containing the toxic protein ricin by using Fourier transform near-infrared (FT-NIR) spectroscopy and multivariate techniques. The method is intended to be a prototype to develop a more general approach to detect food tampering. Measurements were made on an FT-NIR system using a diffuse reflection-integrating sphere. Flours spiked with caffeine, crystalline sugar, and corn meal, 1-20% w/w, were used as test articles to evaluate the methodologies. Food matrices (bleached flour, wheat flour, and blueberry pancake mix) spiked with CBM (0.5-8% w/w) were analyzed. Multiplicative scatter correction transformed partial least-squares regression models, using a specific NIR spectral region, predicted CBM contamination in foods with a standard error of cross-validation of <0.6% and a coefficient of determination (R2) of >94%. Models discriminated between flour samples contaminated with CBM and other protein sources (egg white, soybean meal, tofu, and infant formula). CBM had loading spectra with bands characteristic of amide groups (4880 and 4555 cm-1) and lipids (5800, 5685, 4340, and 4261 cm-1).