The importance of elevated concentrations of oxytocin (OT) during the entire milking period was investigated in seven primiparous dairy cows with inherent disturbed milk ejection and in sixteen healthy cows with disturbed milk ejection induced by placing them in an operating theatre. Disturbance of milk removal in both groups has previously been demonstrated to be exclusively due to central blockage of the expected OT release in response to teat stimulation and milking. However, milk ejection can be induced by exogenous OT. OT (0.2 i.u.) was injected i.v. before milking and 49 +/- 6% of the total milk was removed. When plasma OT decreased, milk flow stopped. In response to a second and third injection of 0.2 i.u. OT, 30 +/- 4 and 7 +/- 2% of the milk were removed respectively. The remaining milk was removed with 10 i.u. OT. The lag time from injection of OT to the start of milk flow was inversely correlated with the amount of milk actually removed in response to the OT injection. If 0.2 i.u. OT was injected during intramammary pressure (IMP) recording, IMP immediately increased to its maximum value. After 2.5 +/- 0.3 min, IMP decreased to an intermediate IMP (between preinjection and maximum IMP). After two additional injections of 0.2 i.u. OT and after injection of 0.5 i.u. OT, IMP increased to a similar maximum. However, after injection of 0.5 i.u. OT, maximum IMP lasted longer (2.9 +/- 0.3 min; P < 0.05) than after injections of 0.2 i.u. If OT was continuously infused (0.15 i.u./min) during milking, milk flow lasted until the udder was completely emptied. IMP increased during OT infusion to a maximum which remained stable until infusion was stopped after 10 min. The same IMP maximum was reached after the first individual OT injection (0.2 i.u.), but when plasma OT decreased towards basal concentrations, milk flow ceased and IMP decreased to an intermediate level. Thus continuously elevated OT concentrations such as those during infusion or during normal milking are necessary for complete milk removal.