BACKGROUND: This study has addressed concerns about possible effects of feeding human infants soy formula milk (SFM). METHODS: This is a feeding study in marmosets, using a mainly co-twin design. From 4–5 until 35–45 days of age, co-twin males were fed by hand with either standard (cow) formula milk (SMA = controls) or with SFM for ~8 h each day (2 h at weekends) and intake related to bodyweight. Blood samples were collected at 18–20 and at 35–45 days of age in 13 sets of co-twins plus two non-twin males per group and, at the later age, seven sets of co-twins were killed and the testes and pituitary gland fixed for cell counts. RESULTS: Weight gain and formula intake were similar in both feeding groups. SMA-fed males had mean testosterone levels of 2.8–3.1 ng/ml, typical of the ‘neonatal testosterone rise’, whereas SFM-fed males exhibited consistently lower mean levels (1.2–2.6 ng/ml); paired comparison in SMA-and SFM-fed co-twins at day 35–45 revealed 53–70% lower levels in 11 of 13 co-twins fed with SFM (P = 0.004). Further evidence for suppression of testosterone levels in SFM-fed males came from comparison of the frequency of low testosterone levels (<0.5 ng/ml). In historical controls aged 35–45 days, two out of 22 values were <0.5 ng/ml, a similar frequency as found in control SMA-fed males (one out of 15 values <0.5 ng/ml). In contrast, 12 out of 15 values for SFM-fed males were <0.5 ng/ml (P < 0.001). There was no consistent relationship between SFM intake/g and testosterone levels. Paradoxically, the mean number of Leydig cells per testis was increased by 74% (P < 0.001) in co-twins fed SFM, when compared with their SMA-fed brothers, whereas no significant changes were found in numbers of Sertoli and germ cells. Because of the lack of gonadotrophin assays, the number of immunopositive LHβ and FSHβ cells in the pituitary gland, and their ratio, were determined but no consistent difference was found between SMA- and SFM-fed twins. CONCLUSIONS: Based on the average isoflavone content of the SFM brand used, intake of isoflavones was estimated at 1.6–3.5 mg/kg/day in the SFM-fed marmosets which is 40–87% of that reported in 4 month human infants fed on a 100% SFM diet. It is therefore considered likely that similar, or larger, effects to those shown here in marmosets may occur in human male infants fed with SFM. Whether the changes described result in longer-term effects is under investigation.