OBJECTIVE To determine the effect of alpha-linolenic acid (ALA) intake (or the dietary linoleic acid [LA]/ALA ratio) on the growth and visual function of term infants. METHODS Normal term infants were assigned randomly and in masked fashion at birth to receive formulas with approximately 16% of total fatty acids as LA and 0.4%, 1.0%, 1.7%, or 3.2% of fatty acids as ALA (LA/ALA ratios of 44, 18.2, 9.7, and 4.8) for the first 4 months of life. The fatty acid pattern of plasma phospholipids was determined shortly after birth and at approximately 21, 60, and 120 days of age. Anthropometric data were obtained at the same times and also at approximately 240 days of age. Transient visual evoked responses (VERs) were measured at approximately 120 and 240 days of age. For comparisons, anthropometric and VER data also were obtained in infants who were exclusively breast-fed for the first 4 months of life. RESULTS Infants who received the formula with 3.2% ALA (LA/ALA ratio, 4.8) had higher plasma concentrations of phospholipid docosahexaenoic acid (DHA) but lower concentrations of arachidonic acid at 21, 60, and 120 days of age. Mean weight of this group at 120 days of age was 760 gm less (p < 0.05) than the mean weight of the group that received the formula with 0.4% ALA (LA/ALA ratio, 44). Despite differences in plasma phospholipid DHA contents among groups, neither VER latency nor amplitude differed significantly among formula groups or between any formula group and age-matched, breast-fed infants. CONCLUSIONS The highest versus the lowest ALA intake (or the lowest vs the highest LA/ALA ratio) resulted in higher plasma phospholipid DHA content from 21 to 120 days of age but was not associated with improved visual function as assessed by transient VER. Moreover, mean body weight of infants who received the highest versus lowest ALA intake was less at 120 days (p < 0.05). These data suggest that the lower LA/ALA ratios currently recommended for infant formulas should not be adopted until the effect of such ratios on growth are evaluated more completely.