Loss of milk quality caused by subclinical infection in dairy sheep has a negative effect on cheese manufacture. As milk from each single animal is not systematically evaluated for somatic cell count, milk from animals with undetected subclinical mastitis often reaches the refrigeration tanks, mixing with normal milk and reducing its technological suitability for cheese manufacture. This study was undertaken to investigate the effect of subclinical mastitis in the coagulation properties of ewe milk using a light backscatter fiber optic sensor. Manchego-type cheese was manufactured using milk from Lacaune and Manchega sheep. Milk from infected and non-infected udders was coagulated and monitored at laboratory scale using both a NIR fiber optic light backscatter sensor and a rheometer. Simultaneously, clotting and cutting time were visually evaluated by an experienced cheesemaker. Optical parameters t(max), t(2max), and t(2min) were highly correlated (0.914 < r < 0.999, P < 0.001) to the visually and rheologically derived clotting and cutting times and with somatic cell counts. It was observed that milk from animals with no udder bacterial infection, irrespectively of the breed, had a quite similar clotting and cutting time. On the other hand, milk from animals having subclinical infection caused by coagulase-negative Staphylococcus had longer coagulation and cutting time. Prediction models using light backscatter parameters alone or in combination with protein/solids concentration were successfully obtained for visually determined clotting and cutting time, rheologically derived gelation and cutting times and for tan delta at cutting with R-2 values ranging from 0.799 to 0.999. Our results suggest that early detection of subclinical mastitis and milk coagulation monitoring using light scatter can diminish the negative impact of mixing milk of infected anirhals, when milk is used for cheese manufacture. (C) 2015 Elsevier Ltd. All rights reserved.