Visible and near infrared spectroscopy using chemometrics was applied for nondestructive quality evaluation of lightly heated shell eggs. Apparent yolk viscosity in the lightly, heated shell eggs was targeted in this study because the yolk viscosity is an important quality attribute related to textural preference and usage suitability to several dishes. The developed partial least squares regression model allows estimation of yolk viscosity in eggs with intact shells. Implementing Martens uncertainty test three times reduced the number of wavelengths and improved the model performance. Moreover, the model predicted yolk viscosity in heated shell eggs with nearly the same yolk viscosity but with profoundly different degrees of egg white gelation. Results indicate that the regression coefficients of the model were insensitive to the egg white gelation status, making the model accurate and robust. (C) 2017 Elsevier Ltd. All rights reserved.