Half-udder comparisons were made using 56 cows for 2 months, in an experiment involving high bacterial challenge, to assess the combined effects of 5 min overmilking and pulsation failure (resulting from the use of shortened teatcup liners) on teat condition and mastitis. Only three new infections were confirmed in over 12500 quarter milkings in quarters milked with control liners (of 148 mm effective length) indicating little or no effect of prolonged overmilking in these quarters. A 3 -5-fold increase in the new infection rate (NIR) based on bacteriological diagnosis alone (P < 0-01), or a 9-fold increase in NIR based on bacteriological diagnosis plus raised cell count and/or iV-acetyl-/?-D-glucosaminidase (P < 0-01), occurred in quarters milked with the same liners shortened to an effective length of 120-130 mm. The mean interval to infection or teat canal colonization (41 v. 60 milkings), and for a quarter infection to be confirmed by other diagnostic tests (45 v. 79 milkings) was significantly less in quarters overmilked with short liners. The results confirm that NIR increases whenever pulsation fails. Overmilking may increase NIR when it is associated with pulsation failure.Experiments in which apparently undesirable machine conditions have been imposed, in conjunction with overmilking, have shown little or no effect of overmilking on the rate of new infections or other indicators of mastitis. Overmilking at high vacuum levels (57-70 kPa) had no effect on new infections, clinical mastitis or udder irritation in heifers (Mochrie et al. 1953;Neave et al. 1962) or in previously uninfected cows (Olney & Mitchell, 1983). Natzke (1978) and Natzke et al. (1978) compared normal milking and overmilking with and without large vacuum fluctuations and concluded that if overmilking is associated with mastitis the effects appear to be small.Milking with short teatcup liners resulted in about twice the new infection rate (NIR) of quarters milked with liners of normal length, but the experimental differences were confined mainly to front quarters (Mein et al. 1983). Although many controlled experiments have failed to show that overmilking seriously affects mastitis (see reviews by Fell, 1964;Dodd & Neave, 1968;Natzke, 1978;Schultze, 1979), it was thought that this interaction in treatment response to pulsation failure between front and hind quarters could be associated with more overmilking on the front quarters. A further, short-term experiment was conducted to assess the combined effects of overmilking and pulsation failure (resulting from the use of shortened teatcup liners) on NIR.