We present preliminary data assessing a previously untried method of deriving estimates of risk from case reports on presumed human teratogens. We postulated that we could take advantage of biases inherent to case reports in order to generate one or more families of temporal curves that could be used to estimate the "true" risk of teratogenic exposure. Using this method (which we refer to as the "case-cumulative method") we found that two agents (parvovirus B19 and isotretinoin) demonstrated a logarithmic decrease in the estimated risk over time, as intuitively expected, while trimethadione and the coumarin derivatives showed a more complex pattern over time. Analysis of estimated risks quoted by reviews and large studies for these four agents showed large variability from estimate to estimate and no discernible temporal pattern. With further analysis of other agents, the case-cumulative method might eventually prove to be useful in teratogen counseling.