PIP: This work explains a classic indirect method for estimating infant mortality and its retrospective tendencies based on census data, and demonstrates an adaptation of the procedure for data taken from Perinatal Clinical History forms. The new procedure is illustrated using data from the Cauca River Valley in Colombia and from Uruguay. The Perinatal Clinical History forms were introduced in 1983 by the Latin American Center for Perinatology as a simple and inexpensive record containing the minimum information needed for adequate prenatal care. Indirect estimates of the type developed by Brass and others based on a census or survey require data on the mother's age, previous live births and surviving and decreased children. The data are 1st divided into 7 5-year cohorts by maternal age and the numbers of mothers, previous births, and surviving and decreased children are calculated for each group. The proportion of children deceased is calculated for each maternal cohort. Although this proportion constitutes a form of mortality measure in itself, it has the limitation of not being a conventional indicator and it also refers to the ages of the mothers rather than of the children. Brass demonstrated that a relationship exists between the proportion of deceased children and the probabilities of dying from birth through an exact age X. An appendix to this work presents the Brass method for estimating with greater precision the probabilities of dying before an exact age based on the proportions of children dying, mean parities, and the West model life table. The Perinatal Clinical History furnishes information on women who constitute not a random sample, but a sample of women who have just given birth and who therefore have not given birth and who therefore have not given birth for approximately the past 30 months (the average birth interval). The procedure proposed by Brass and Macrae allows a progression from infant mortality estimates based on the experience of women who have just given birth to unconditional estimates that would apply to the entire population. An estimate of unconditional mean parity is obtained by adding to the average previous parity of the sample women (not counting the current birth) 1/2 the number of women in the sample. An additional adjustment is necessary to reflect inclusion of all women, and not just mothers, in the estimate. Information on the total number of women of each age group should be obtained from a source external to the Perinatal Clinical History. The mortality of the additional infants is assumed to be identical to that of infants already included. Once these adjustments are made, the estimation procedure is identical to that using census data. Estimates obtained from census data and from the Perinatal Clinical History cards in the Cauca River Valley and in Uruguay were reasonably close, suggesting that the technique would be useful for assessing and monitoring infant mortality trends in countries lacking reliable vital statistics.^ieng