Given a national census and a life table we may wish to infer the birth rate and hence the rate of natural increase. The model for making the inference here presented specifies the form of the preceding births and an intuitively selected life table if death statistics are not available. A signal on the appropriateness of the model is provided by (a) the resemblance of the age distribution derived from the model to the observed age distribution, and (b) the closeness of the inferred death rate to the observed death rate which has been implicitly fed into the computation; when either of these disparities is too great no result is printed out. Three other ways of obtaining an estimate of births are presented, each with its own characteristic properties. When the four methods agree with one another they can be applied to provide a birth rate in default of birth statistics, or to suggest a correction for birth statistics which are deficient. A table showing results on the four methods for 25 cases enables the reader to judge the practical usefulness of the system.