A new method is proposed for the surveillance of Down's syndrome among newborn. Despite the strong dependence of overall risk of Down's syndrome on maternal age, it has been suggested that an environmentally induced increase in risk may be additive over all maternal ages. The surveillance method introduced here is specifically designed to detect such changes. The method is based on registry data for successive periods for a given population. It is assumed that the number of Down's syndrome cases as well as the total number of births are known in all maternal age groups. Tables of average run lengths until an alarm (ARLs) are calculated for a total sample size of 14,500 in each period, the approximate number of births in a three-month period in Norway. Comparison with the Poisson cusum shows that the new surveillance method can detect moderate additive increases significantly faster. Applied retrospectively to quarterly data from the Medical Birth Registry of Norway for 1978-89, the proposed method was close to an alarm in 1985 and actually signalled a strong alarm in 1986, reflecting a previously reported increase in risk in this period. The cusum method was not so sensitive to the aberration in Down's syndrome risks in 1985 and 1986.