Food emulsions with high amounts of unsaturated fats, such as mayonnaise, are prone to lipid oxidation. In the food industry, typically accelerated shelf life tests are applied to assess the oxidative stability of different formulations. Here, the appearance of aldehydes at the so-called onset time, typically weeks, is considered a measure for oxidative stability of food emulsions, such as mayonnaise. To enable earlier assessment of compromised shelf-life, a predictive model for volatile off-flavor generation is developed. The model is based on the formation kinetics of hydroperoxides, which are early oxidation products and precursors of volatile aldehydes, responsible for off-flavor. Under accelerated shelf-life conditions (50 °C), hydroperoxide (LOOH) concentration over time shows a sigmoidal curvature followed by an acceleration phase that occurs at a LOOH-concentration between 38–50 mmol/kg, here interpreted as a critical LOOH concentration (CCLOOH). We hypothesize that the time at which CCLOOH was reached is related to the onset of aldehyde generation and that the characterization of the LOOH-generation curvature could be based on reaction kinetics in the first days. These hypotheses are tested using semi-empirical models to describe the autocatalytic character of hydroperoxide formation in combination with the CCLOOH. The Foubert function is selected as best describing the LOOH-curvature and is hence used to accurately predict onset of aldehyde generation, in most cases within several days of shelf-life. Furthermore, we find that the defining parameters of this model could be used to recognize antioxidant mechanisms at play.