The biases involved in the collection of data for the Surveys of Infant and Child Mortality and the Surveys of Infant Mortality in the Sahel were largely foreseen from the outset. This work does not address biases resulting from exclusion from the sample of women giving birth at home or outside the city or women leaving the hospital before being interviewed by survey personnel; an estimated 12.4% of newborns in Yaounde were not represented for these reasons. Biases resulting from the progressive thinning of the sample due to temporary absence of the mother, emigration, or other cause are more fully explored. There is every reason to believe that modifications to the sample resulting from departures of mothers are qualitative as well as quantitative, and that the explanations of their departures involve the death of their child. The resulting biases can be evaluated and corrected using data contained in the survey. A proposed formula for calculating mortality quotients for different rounds of the survey is correct in its assessment of probabilities but ambiguous in failing to take into account changes in the sample from 1 round to another. The proposed solution is to calculate for each interval between rounds a series of mortality quotients for homogeneous subpopulations identified on the basis of the variables most directly related to both departures from the sample and mortality. To obtain a mortality quotient applicable to the entire population, the quotients should be reweighted according to the distribution of subgroups that would have been observed in the sample in the absence of departures from observation for other reasons than mortality. The variables needed to divide the population into homogeneous categories should be taken from the basic questionnaire completed in the hospital. They cannot be definitely selected until processing of data has started, so that their correlation with the risks of mortality and loss to the sample have been demonstrated. This procedure has the advantages of corresponding to 1 of the survey objectives, exploring social determinants of mortality, and of improving the precision of estimates. Other methodological questions explored in this document include treatment of cases returned to the sample after temporary absences, which should differ according to the manner in which they were returned to the sample, adoption of different age divisions than the 4-month intervals resulting from the survey methodology, and assessing seasonal factors in infant mortality.