Clusters of adverse reproductive outcomes are reported with increasing freq
uency to public health authorities. Most are random events and only a very
small proportion is likely due to identifiable environmental agents. Often,
a preliminary study confirms the existence of a spatial and temporal exces
s but no biologically plausible cause is found. These cases require a follo
w-up of incidence to identify any continuing excess. A conceptually simple
'fixed window' technique of follow up is proposed. The power of the test is
mainly influenced by the magnitude of the increase in rate, the number of
adverse outcomes selected for observation and the acceptable false alarm ra
te. This technique has several advantages compared with other currently use
d methods. From a public health point of view, two important factors are to
be considered in setting the parameters of the test: the delay in providin
g an answer for the community and the availability of resources for conduct
ing environmental investigations.