Condensed tannins have been considered to be important inducible defenses against mammalian herbivory. We tested for differences in condensed tannin defenses in Acacia drepanolobium in Kenya over two years among different large mammalian herbivore treatments [total exclusion, antelope only, and megaherbivore (elephants and giraffes) + antelope] and with four different ant symbiont species on the trees. We predicted that (1) condensed tannin concentrations would be lowest in the mammal treatment with the lowest level of herbivory (total exclusion), (2) trees occupied by mutualist ants that protect the trees most aggressively would have lower levels of tannins, and (3) if chemical defense production is costly, there would be a trade-off between tannin concentrations, growth, and mechanical defenses. Mean tannin concentrations increased from total exclusion treatments to wildlife-only treatments to megaherbivore + antelope treatments. In 1997, condensed tannin concentrations were significantly lower in trees occupied by the ant Crematogaster nigriceps, the only ant species that actively removed axillary buds. Contrary to our prediction, trees occupied by ant species that protect the trees more aggressively against mammalian herbivores did not have lower overall levels of condensed tannins. There was no consistent evidence of a trade-off between tannin concentrations and growth rate, but there was a positive correlation between mean thorn length and mean tannin concentrations across species of ant inhabitants and across herbivore treatments in 1997. Contrary to our expectation, trees had higher tannin concentrations in the upper parts of the canopy where there is little herbivory by mammals.