Three types of liner (wide bore tapered, wide bore and narrow bore) were compared in a change-over design experiment comprising three 3-d periods and using six lactating Holstein Friesian cows with a wide range in milk flow rate and teat dimensions. Vacuum levels measured within the liner showed a characteristic pattern of three distinct phases: 1, a mouthpiece chamber (MPC) vacuum that fluctuated with vacuum beneath the teat; 2, a stable MPC vacuum that coincided with the period of peak milk flow and 3, a MPC vacuum that fluctuated with pressure changes in the pulsation chamber and coincided with the period of no or reduced flow. Of the three liners examined, the Gascoigne 1030A (wide bore tapered) induced the highest MPC vacuum (29·4 ± 2·23 kPa, mean ± s.e.m), significantly higher (P < 0·01) than the Alfa-Laval 960000–01 wide bore (23·5 ± 2·32), regardless of teat size and milk flow rate. The Alfa-Laval 7275737–01 (narrow bore) gave a significantly lower value (16·5 ± 1·96) than either of the others (P < 0·001). It was not possible to link absolutely raised MPC vacuum to cow discomfort and teat trauma but circumstantial evidence encourages this view. Narrow teats in combination with slow flow rate appeared to exacerbate MPC vacuum and, because milking time for such animals was protracted, teats were exposed for longer periods.