Subclinical mastitis is a frequent and economically important disease in cattle. Since the milk appears macroscopically normal the diagnosis depends on indirect parameters. In automatic milking systems ideally the measurements have to be automatized and completed during milking process. Oxygen concentration of the milk is a parameter which has not been used widely to date. Oxygen is consumed by cells in the milk, hence an increased number of cells results in decreased oxygen concentration. The main objectives of the study were to study the association of milk oxygen concentration with the number of cells and the electric conductivity and to assess the feasibility of oxygen concentration for detection of subclinical mastitis. The study was performed on five dairy farms using 690 macroscopically normal quarter milk samples. Oxygen concentrations (OC), somatic cell count (SCC) and electric conductivity (EC) were measured. The associations between the parameters were calculated and the diagnostic value of OC for detection of subclinical mastitis was estimated. Significant correlations were found between OC and SCC (rs =-0.28) and between OC and EC (rs = -0.20). SCC of the samples varied between 1,000 and 21,602,000 cells/ml. Out of the 690 milk samples 404 had SCC 200,000 cells/ml (subclinical mastitis). OC decreased whereas EC increased significantly with increasing SCC (P < 0.001). The threshold for OC to diagnose subclinical mastitis was 5.22% (sensitivity 84%, specificity 46%). The area under the curve describing the diagnostic value of OC was 0.72 (confidence interval 0.68-0.76). In conclusion OC has potential to be used as parameter to detect subclinical mastitis in dairy cattle. Currently the application is not suitable for routine use. Further research is necessary to improve measurement technology and diagnostic value of the parameter.