Many studies have failed to detect costs of defense and some have even found a positive correlation between growth and the concentrations of chemical defenses. These studies contradict the theoretical assumption that anti-herbivore defenses are costly-produced at the expense of growth and/or reproduction. Costs, however, may be transient and therefore difficult to detect. Here we tested the hypothesis that costs of defense would be pronounced early in development when root growth is prioritized (high percent root allocation), but not later in development. To test this hypothesis, we grew F(2) hybrid willow seedlings from five different families, and harvested cohorts of even-aged seedlings after 6, 7, 8 and 9 weeks of growth. Seedlings were divided into root and shoot tissue and shoots were analyzed for phenolics (condensed tannins and phenolic glycosides). We found evidence for transient costs of defense. The concentrations of phenolics were negatively correlated with total biomass, shoot biomass, and the proportion of biomass allocated to roots in week 6. After week 6, however, the concentrations of phenolics were positively correlated with shoot biomass and total biomass, while phenolics were uncorrelated with the proportion of biomass allocated to roots. These results, the first ever, to our knowledge, with woody plants, suggest that costs of defense were transient; specifically, costs were found in early development, when root establishment was a priority. Our findings suggest that studies should focus more on trade-offs early in plant development.