The objectives of the study were: to evaluate the interaction between visual appearance and California mastitis test (CMT) score of the foremilk in relation to the cell count of the milk; to evaluate the consequences of sorting milk according to these criteria; and to explore whether visual appearance and CMT score of foremilk depended on the time interval between milkings. Measuring somatic cell count (SCC) in composite milk only and discarding milk above certain thresholds will not ensure that milk from all cows with visually abnormal foremilk is withheld from delivery. Low thresholds of SCC will reduce the frequency of cows with abnormal milk but increase the discarding of milk from cows with visually normal foremilk. CMT score of foremilk differentiated better between cows with high and low SCC in composite milk than visual inspection of foremilk. CMT scores of foremilk decreased with increasing interval between milkings within cow, whereas the visual appearance was independent of the interval. We propose that visual appearance of the foremilk should be kept as a criterion for sorting milk at time of milking. For test purposes, the use of visual appearance of foremilk for differentiation between normal and abnormal milk has to be done on multiple milkings. Additionally, CMT scoring of foremilk improves correct classification of normal and abnormal quarters and especially when including data from the previous milking.