OBJECTIVE To assess the ability of a hand-held device to differentiate between infected and noninfected bovine mammary glands according to the electrical resistance of milk, under Australian conditions. METHODS A cross-sectional study. METHODS Milk samples were collected from 236 quarters of 60 cows selected from a commercial dairy herd with a high prevalence of mastitis. The true infection status of these quarters was determined using bacteriology. Various methods were used in an attempt to relate the electrical resistance of milk from each quarter to the presence or absence of infection in that quarter. RESULTS Although the electrical resistance of milk from infected quarters was generally lower than that of noninfected quarters, the overlap of readings between the two populations limited the ability of this device to indicate accurately whether a quarter was infected. The use of methods comparing the readings from each of the four quarters of a single cow did not allow the reliable detection of infected cows. CONCLUSIONS Although this device may have some practical advantages in comparison with some other methods of diagnosing subclinical mastitis, the predictive value of the method was generally poor.