Multiples of medians of serum markers are assumed to be independent of gestational age: every algorithm used for Down's syndrome risk evaluation is based on this hypothesis. However, our former observations suggested that multiples of medians of human chorionic gonadotrophin in Down's syndrome are dependent on gestatational age. Furthermore, observations on 84 Down's syndrome cases confirmed that human chorionic gonadotrophin multiples of medians in samples drawn at 15-17 weeks are approximately 10% lower than in samples drawn at 18-21 weeks, thus showing that the human chorionic gonadotrophin concentration decreases about 10% less than expected. The control group comprised 554 women with two blood samples and normal human chorionic gonadotrophin at first sampling. A further group of 532 women with multiples of medians at first sampling > 1.8 was examined with the aim of excluding an association between the human chorionic gonadotrophin trend in Down's syndrome and high starting values. The trend is peculiar to human chorionic gonadotrophin in Down's syndrome pregnancies and may help to explain the increase in detection rate with gestational age. Based on these findings, screening can be optimized, thus improving performance.