In this study, visible-near-infrared (VNIR) spectroscopy along with proximate analysis was employed to develop and validate partial least square regression (PLSR) models for predicting cholesterol, fat, calories, and moisture content of fresh and cooked ground beef patties. The refined PLSR model utilized 51 bands from the NIR, green and red regions for raw patties, and 50 bands from NIR and blue regions for cooked patties. The PLSR models resulted in high validation R 2 values of 0.72-0.94, with the standard error of prediction varying from 1% to 8.9%. However, the residual predictive deviation (RPD) was acceptable only for calories and fat in both raw and cooked patties, and moisture in raw patties. This study indicated that VNIR spectroscopy in combination with PLSR modeling can be used for screening fat and calories in both raw and cooked beef patties, and may hold the potential for cholesterol screening. (c) 2008 Elsevier Ltd. All rights reserved.