Sudan-1 has been used for coloring food. However, recent alarms worldwide about the carcinogenic and mutagenic properties of azo-compounds have led to concerns over their human consumption. In the U.K. in 2005, over 570 products were found to be contaminated with the azo dye Sudan-1 and this and the health risks associated with this dye resulted in the subsequent international ban of this additive in all foodstuff, at all levels, relating to human consumption. These incidents have also necessitated the need for high throughput low cost reliable approaches for the detection and quantification of food contaminated by such azo compounds. While there are a small number of analytical techniques that can be considered portable, many lack sensitivity. By contrast, we show that employing a portable Raman spectrometer, using surface enhanced Raman scattering (SERS), can provide good sensitivity, such that Sudan-1 can be quantified in a complex food matrix reliably over the range of 10 -3 to 10 -4 mol L -1. We also demonstrate that a variety of multivariate approaches including principal components analysis (PCA), partial least-squares (PLS) regression, artificial neural networks (ANNs), and support vector regression (SVR) can be employed for the chemical analysis of this dye in a quantitative manner. Compared to the commonly used univariate approaches, where the area under a single band in assessed, the advantage of using multivariate approaches is that these algorithms can analyze the full spectra directly and the laborious task of selecting and integrating marker appropriate quantitative spectral bands can be avoided thus greatly simplifying and speeding up data analysis.