The evolution of increased competitive ability (EICA) hypothesis proposes that invasive species evolve decreased defense and increased competitive ability following natural enemy release. Previous tests of EICA examined the result of evolution by comparing individuals from home and introduced ranges, but no previous study of this hypothesis has examined the process of evolution by analyzing patterns of selection. On the basis of EICA, there should be selection for competitive ability without herbivores and selection for defense with herbivores. Selection on competitive ability should be stronger for genotypes accustomed to herbivores (home range genotypes), and selection on defense should be stronger for genotypes unaccustomed to herbivores (introduced range genotypes). Using a field experiment, we tested these hypotheses for the invasive plant Melaleuca quinquenervia. There was a negative genetic correlation between resistance and growth, indicating a trade-off. However, selection for stem elongation (an indicator of competitive ability) was always positive, and selection on resistance was always negative and did not depend on genotype source or the presence of herbivores. The patterns of selection found in this study contrast with predictions from EICA and accurately predict the lack of evolutionary change in growth and resistance following the introduction of this species from Australia to Florida.