A total of twelve lactating Jersey cows were used in a 5-week experiment to determine the effects of severe feed restriction on the permeability of mammary gland cell tight junctions (TJs) and its effects on milk stability to the alcohol test. During the first 2 weeks, cows were managed and fed together and received the same diet according to their nutritional requirements (full diet: 15 kg of sugar cane silage; 5.8 kg of alfalfa hay; 0.16 kg of mineral salt and 6.2 kg of concentrate). In the 3rd week, animals were distributed into two groups of six cows each. One group received the full diet and the other a restricted diet (50% of the full diet). In the 4th and 5th weeks, all animals received the full diet again. Milk composition and other attributes, such as titratable acidity, ethanol stability, pH, density and somatic cell count (SCC) were evaluated. Cortisol levels indicated the stress condition of the cows. Plasma lactose and milk sodium were measured to assess mammary TJ leakiness. Principal factor analysis (PFA) showed that the first two principal factors (PFs) contributed with 44.47% and 20.57% of the total variance in the experiment and, as feeding levels increased, milk stability to the ethanol test became higher and plasma lactose levels decreased, which indicates lower permeability of the mammary gland cell TJ. Correspondence analyses were consistent with PFA and also showed that lower feeding levels were related to reduced milk stability, high plasma lactose, high sodium in milk, low milk lactose (another parameter used to assess TJ permeability) and higher cortisol levels, indicating the stress to which animals were submitted. All observations were grouped in three clusters, with some of the above-mentioned patterns. Feeding restriction was associated with higher permeability of TJ, decreasing milk stability to the ethanol test.