This work investigates how the temperature dependence of oil viscosity varies with the type of frying oil,
initial oil temperature, frying load and number of frying repetitions, during potato deep fat frying. Viscosity
is measured in small temperature increments over a broad temperature range. A non-linear model is
proposed which gives more statistically significant results than other known models in describing the
temperature dependence of viscosity. Multiple quasi-linear regression analysis is applied to derive an
expression that predicts oil viscosity from important frying conditions such as oil type, frying load, average
oil temperature and frying time. For applications where only the initial oil temperature is known, a
correlation is proposed for the prediction of the average oil temperature from the initial oil temperature
and other frying conditions. The overall accuracy of the model in predicting the temperature dependence
of viscosity on the examined frying conditions is better than 95.0%.