Seven variables--electrical conductivity (EC), somatic cell count (SCC), N-acetyl-beta-D-glucosaminidase (NAGase), lactose, protein, fat and pH--were compared in four quarter milk fractions (MF1: strict foremilk; MF2: first 12-15 ml foremilk; MF3: subsequent 40-45 ml milk; MF4: strippings) and in one cow composite milk sample (CC) per cow. The study used 142 quarters from 37 lactating cows of the German Black Pied breed. To rule out any possible effect due to management, animal physiology and analytical procedures, the collection and processing of milk samples from each cow was repeated for three consecutive days, and the means of 3-d values were used. All variables were affected significantly by milk fraction and udder health. Compared with foremilk, EC, lactose and protein levels in strippings decreased, while SCC, NAGase and fat increased. The pH of foremilk and strippings did not differ significantly in healthy or in mastitic quarters. The difference between MF1 and MF2 was significant for EC in mastitic quarters, and for SCC in healthy quarters only. In general, mastitis resulted in a significant increase in EC, SCC, NAGase and protein but in a decrease in lactose and fat contents of milk in one or more of the milk fractions studied. Comparison of cow composite milk samples from healthy and mastitic cows revealed the significance (P < 0.01) of udder health for EC, SCC and lactose. Of the different parameters that can distinguish between healthy and mastitic quarters or cows, EC could be used to classify 76% of quarters and 73% of cows correctly, while the lactose content permitted correct identification of 81% of quarters and 76% of cows. NAGase and pH could be used to determine the status of 73% and 61% of quarters, respectively. In general, the correlation observed in strippings was higher than in foremilk for almost all the variables studied. Surprisingly, EC, SCC, NAGase and lactose in milk from healthy quarters of mastitic cows (with at least one mastitic quarter) differed significantly (P < 0.05) from those from healthy quarters of cows with all four healthy quarters, indicating an inconsistent effect of mastitic quarters on neighbouring healthy quarters (quarter interdependence).