Whole genome duplications (WGDs) have been hypothesized to be responsible for major transitions in evolution. However, the effects of WGD and subsequent gene loss on cellular behavior and metabolism are still poorly understood. Here we develop a genome scale evolutionary model to study the dynamics of gene loss and metabolic adaptation after WGD. Using the metabolic network of Saccharomyces cerevisiae as an example, we primarily study the outcome of WGD on yeast as it currently is. However, similar results were obtained using a recontructed hypothetical metabolic network of the pre-WGD ancestor. We show that the retention of genes in duplicate in the model, corresponds nicely with those retained in duplicate after the ancestral WGD in S. cerevisiae. Also, we observe that transporter and glycolytic genes have a higher probability to be retained in duplicate after WGD and subsequent gene loss, both in the model as in S. cerevisiae, which leads to an increase in glycolytic flux after WGD. Furthermore, the model shows that WGD leads to better adaptation than small-scale duplications, in environments for which duplication of a whole pathway instead of single reactions is needed to increase fitness. This is indeed the case for adaptation to high glucose levels. Thus, our model confirms the hypothesis that WGD has been important in the adaptation of yeast to the new, glucose-rich environment that arose after the appearance of angiosperms. Moreover, the model shows that WGD is almost always detrimental on the short term in environments to which the lineage is preadapted, but can have immediate fitness benefits in "new" environments. This explains why WGD, while pivotal in the evolution of many lineages and an apparent "easy" genetic operator, occurs relatively rarely.