Despite the extensive research performed during the last decades, the multifactorial mechanism responsible for white wine protein haze formation is not fully characterized. A model is proposed, which is essentially based on two postulates: the experimental identification of sulfur dioxide as the non-proteinaceous factor, and the inference from reliable data available in the literature of the dynamic chemistry played by wine protein sulfhydryl groups. Unlike other reducing agents, addition of SO2 to must/wine upon heating cleaves intraprotein disulfide bonds, hinders thiol-disulfide exchange during protein interactions, and leads to formation of novel interprotein disulfide bonds. These bonds are ultimately responsible for wine protein aggregation following a nucleation-growth kinetic model, as shown by Dynamic Light Scattering experiments. The model was tested in wine model solution (using total and fractionated wine proteins) and validated under real wine conditions. The results achieved may open the way to develop techniques that will find wide application in the wine industry.