Mastitis is the most frequent and costly disease in dairy herds, as it negatively affects yield and milk quality. The presence of clinical mastitis is quite easy to asses, whereas the diagnosis of the subclinical form can be more difficult and requires laboratory assays. Somatic cell count (SCC) is widely used as a rapid and low-cost indicator of mastitis, even if is not useful in discriminating between the clinical and subclinical form. As amyloid A has been investigated as a marker of mastitis, the aim of this study was to assess the potential value of measuring amyloid A in serum and milk and the correlation with SCC in the diagnosis of subclinical mastitis. The reliability of two different ELISA kits for the measurement of amyloid A in milk was also tested. During a 1-month trial period, 21 cows were assigned to three experimental groups according to their health status: 6 cows with clinical mastitis (CM), 10 cows with subclinical mastitis (SM) and 5 healthy cows (HE). Amyloid A was measured both in serum (SAA) and in quarter milk samples (mAA) with a serum ELISA kit, and in quarter milk samples (MAA) with a milk ELISA kit. SCC, total microbial count (TMC) and bacterial examination of the milk were also carried out. After a log transformation, the data were submitted to ANOVA and linear regression. TMC was significantly higher in cows with clinical mastitis, while no differences were observed between the other two experimental groups. SCC and MAA levels were significantly different among the three groups. mAA concentrations were similar between cows with subclinical and clinical mastitis, and SAA was not affected by mastitis. A significant correlation between SCC and MAA or mAA was detected, while no correlation was recorded between SAA and mAA. A close relationship between MAA and mAA was noticeable even at low concentrations, suggesting MAA as a potential physiological marker of subclinical mastitis.