The objective of this study was to develop models to predict SCC (somatic cell count) in unhomogenized milk by visible and near-infrared (NIR) spectroscopic technique. Total of 100 milk samples were collected from dairy farms and preserved to minimize propagation of bacteria cells during transportation. Reductive reagents such as methyl red, methylene blue, bromcresol purple, phenol red and resazurin were added to milk samples, and then colors of milk were changed based on SCC of milk. For optimal reductive reagents, reaction time was controlled at 3 level of reaction time. A spectrophotometer was used to measure reflectance spectra from milk samples. The partial least square (PLS) models were developed to predict SCC of unhomogenized milk. The PLS results showed that milk samples with reductive reagents had a good correlation between predicted and measured SCC at 5 minutes of reaction time in the visible range. The PLS models with resazurin reagent had the best performance in $400{sim}600;nm$. The prediction results of milk samples with resazurin had 0.86 of correlation coefficient and 14,184 cell/mL of SEP.