Using standard populations like "standard-primipara" (normal pregnancy, singleton term delivery and cephalic presentation) and "caesarean secundapara" (previous caesarean section and second birth) as the basis for interunit comparisons of maternity care will control for differences in casemix that may be seen at different units, thereby increasing the validity of comparisons. Focusing on clinically meaningful subsets of the population may have the additional benefit of clarifying the relationship between everyday clinical decision making, and the statistics from medical birth registration. Birth registry data from Rigshospitalet, Hvidovre Hospital and Herning Centralsygehus 1993-1994 have been used to illustrate the association between local quality improvement activities, on the one hand, and rates of interventions and foetal outcome in "standard-primipara" on the other.