Objectives: To characterize an early trait in the BMI-for-age curve, the infant BMI peak. Methods: BMI-for-age curves were produced for 747 non-Hispanic, white Fels Longitudinal Study participants, from which individual age (AgePeak) and BMI (BMIPeak) at maximum infant BMI were estimated. Multivariable general linear regression was used to examine the effects of sex and birth year cohort (1929-1950, 1951-1970, and 1971-2010) on AgePeak and BMIPeak, with associations between BMIPeak and concurrent sum of four skinfold thicknesses assessed in a subsample (N=155). Heritability (h2) of AgePeak and BMIPeak was estimated using maximum-likelihood variance components analysis. Results: AgePeak occurred at 9 months of age in both sexes, but BMIPeak was 0.4 kg/m2 higher for boys than for girls (P-value<0.001). Infants born between 1971 and 2010 experienced a 1.5 month earlier AgePeak and a 0.35 kg/m2 lower BMIPeak than infants born between 1929 and 1950 (P-values<0.001). Skinfold thickness explained 37% of the variance in BMIPeak in boys and 20% of the variance in girls (p-values<0.001). AgePeak and BMIPeak were significantly heritable (h2=0.54 and 0.75, respectively). Conclusions: Both AgePeak and BMIPeak decreased over successive birth year cohorts in the Fels Longitudinal Study. Despite a positive association of BMIPeak with concurrent adiposity, AgePeak appears to occur later than does the well-documented peak in infant fat mass and BMIPeak does not capture known sex differences in infant adiposity. Strong heritability of these infant BMI traits suggests investigation of genetic control, and validation of their relationship to body composition is greatly needed.