The validity of mothers' reporting of the breast feeding history of their children more than 20 years after their birth was studied in a Jerusalem population. Among 74 study subjects, duration of breast feeding as ascertained from an interview was well correlated with that recorded in mother and child health clinic charts. Concordance was of similar magnitude in subgroups of ethnicity, mother's education, family size, and sex of child. Duration of breast feeding in 101 youngsters was inversely associated with plasma cholesterol in 17 year old girls, though not in boys, which was statistically significant on univariate analysis and of borderline significance on multivariable analysis. Among 17 year old boys, though not in girls, a statistically significant inverse association for plasma triglyceride was apparent on multivariable analysis. Reported breast feeding history derived from interview of mothers may be a useful instrument for study of possible long term effects of breast feeding in their adolescent or adult progeny.