Abstract. The perinatal mortality rate at the Danderyd Hospital during 1968–70 (period I, 9 801 births) was 16 per thousand total births. In spite of inadequate resources for fetal heart rate monitoring, intrapartum mortality formed only 11% of perinatal mortality. Antepartum death constituted no less than 42% of the perinatal mortality. The majority of antepartum deaths occurred before admission to hospital and more than 50% were without congenital abnormality. In spite of being intensive the conventional antenatal care was therefore regarded as poor in predicting retarded, or accelerated fetal growth and twin pregnancies. Therefore the gravidogram system was designed for the supervision of pregnancy by means of graphic comparisons between changes in maternal Symphysis-fundus (SF) distance, girth, weight and blood pressure, and known normal values. Increases of maternal weight, girth and SF-distances in 100 uncomplicated term pregnancies (50 nulliparae and 50 multiparae) resulting in normal for dates infants (NFD=mean weight ± 1 S.D. for gestational week in question) are presented as means ± 2 S.D. SF-distance had the smallest coefficient of variation and was therefore selected as an indirect indicator of fetal growth. SF-curves for 50 large for dates (LFD) and 50 small for dates (SFD) infants were constructed. Already by the 20th week of pregnancy the SF-growth curve for LFD-infants differed significantly from that of NFD-infants (p < 0.0l). The SF-growth curve for SFD-infants differed significantly from that of NFD-infants by the 28th week of gestation (p < 0.01). Antenatal screening was performed by SF-tape measurements (mean ± 1 S.D.) in a consecutive series of 428 women delivered between 37th and 42nd week of pregnancy. When SF-growth charts were normal 80–84% of infants were NFD. The number of infants detected in this way comprised 64% of all NFD-infants, which is close to statistical expectation. When SF-growth charts were above the normal (> mean +1 S.D.) 65% of all LFD-infants were correctly predicted. Similarly when SF-growth charts were below normal (< mean −1 S.D.), static ordeclining, 75% of all SFD-infants were predicted. SF-growth charts were analysed in 69 twin pregnancies and the clinical diagnosis improved from the wellknown 50% level to 86%. If SF-values greater than the mean +2 S.D. (NFD-singleton) were considered as suspicious of twin pregnancy this suspicion could be confirmed in all but one case (99%). Average time for suspicion of a twin pregnancy was 24±4 (1 S.D.) gestational weeks. Human placental lactogen (HPL) and urinary estriol (U-estriol) were determined 2–3 times weekly during third trimester in pregnancies resulting in LFD-infants (n = 19) and SFD-infants (n = 29). SF-measurements were superior both to HPL and U-estriol in detecting accelerated and retarded fetal growth. After the introduction of the gravidogram there was a prompt and persistent fall in the uncorrected perinatal mortality rate to 8.0 per thousand total births in the last 12 800 births. A normal SF-growth curve implies normal fetal growth and the risk of intrauterine death from fetal growth retardation is practically nil. Biochemical and sonographic fetal supervision can therefore be limited to cases where the SF-growth chart deviates from normal. The perinatal mortality in Sweden and at the Danderyd Hospital during the period 1961–1975 are compared and discussed.