The objective of the present work was to identify the compositional parameters of raw milk that affected ethanol stability at natural pH when natural milk conditions were not modified. Heat stability, measured as coagulation time (CT), was included in the analysis to verify relation to alcohol test. Statistical models were proposed for alcohol and heat (CT) stabilities. Milk samples of good hygienic quality from dairy farms were classified in two groups according to their alcohol stability. Unstable samples to ethanol (72%, v/v) presented lower values of pH, somatic cells count, casein and non-fat-solids relative to ethanol stable samples (ethanol at 78%, v/v or more); whereas freezing point, chloride, sodium and potassium concentrations were higher in the unstable group. Logistic regression and multiple regression were applied to modelling alcohol and heat stability behaviour respectively. Chloride, potassium, ionic calcium and somatic cell count were included in the alcohol regression model, whereas calcium, phosphorous, urea, pH and ionic calcium were part of CT model. Ionic calcium was the only measured variable that contributed to both models; however coagulation time was noted to be more sensitive to ionic calcium than alcohol. The relation between ionic strength and casein was found to contribute to the alcohol model but not to the CT model. However, the interaction calcium plus magnesium plus phosphorous and casein contributed only to CT model.