Fertility trends estimated alternately from birth histories and own children method are compared for eight developing countries in which the World Fertility Survey was conducted. Principle hypotheses are that fertility trends estimated by the two approaches suffer from similar errors in the reporting of women's and children's ages, and that these errors are more severe in estimates derived from own children than in estimates derived from birth histories. The hypotheses are confirmed in four of the eight countries. Potter's hypothesis about misplacement of events toward the survey date, which assumes bunching of births five to ten years before the survey and accurate reporting of births during first five years or so immediately preceding the survey, resulting in a spurious estimated fertility decline, does not receive much support from these data. Some of the estimated fertility declines do indeed seem spurious, but Potter's explanation of them seems inconsistent with the data. In some countries, patterns of age misreporting involving upward rounding of children's ages provide an equally plausible explanation that is more consistent with the data. (Author)