Fatty acid (FA) profile is a crucial factor in defining milk properties and quality. Increasing attention is placed on those FA having a positive role on human health, such as polyunsaturated FA (PUFA), oleic acid, vaccenic acid and conjugated linoleic acid (CLA). In the present work, the FA profile obtained from gas-chromatographic analysis of 703 individual milk samples (289 cows, 214 buffaloes and 200 ewes), were used to evaluate the suitability of principal component (PC) analysis (PCA) to investigate differences in milk FA profile among these species. Thirty-six FA were analyzed by PCA. The first and second PC (PC1 and PC2) accounted for 51% of the total variability (30%, 21%, for PC1, PC2, respectively). Elements of eigenvector of PC1 were positively associated with few FA, including C16:0 (0.50), the most abundant FA in milk of the 3 species (~29.3% of FA). Among those negatively correlated with PC1, some relevant FA for human health were retrieved: C18:1t11 (−0.75), CLAc9t11 (−0.79), C18:3n-3 (−0.76), EPA (−0.81) and DPA (−0.89). Negative correlations with PC2 were observed, in general, for medium chain FA (e.g., C15:0, C16:0), whereas long chain FA (e.g., C18:1 isomers, C18:2n-6 and other PUFA) exhibited positive values. Score values of PC1 and PC2 allowed the description of the relationship among animals based exclusively on the milk FA profile, revealing a clear discrimination of the 3 species: PC1 discriminated ewes milk from that of cows and buffaloes, whereas, PC2 was able to discriminate cows from buffaloes, with ewes showing intermediate values. In general, the use of PCA evidenced as a great part of variance of FA profile of ruminants (PC1) is related to the distinction between small or large ruminants, also evidencing higher amount of beneficial FA for ewes. An important part of variance (PC2) indicates as ewes milk FA has intermediate characteristics compared with cow and buffalo milk.