Relaxor ferroelectrics have been a focus of intense attention due to their fascinating physical properties. Their diffuse phase transitions have been explained by the polar nanoregion model. Nevertheless, fundamental characterization of structure and dynamics in relaxors is still a long-standing challenge. Better scientific understanding of the microscopic origins of relaxor behavior is also required to improve efficiencies of relaxor based devices. Our molecular dynamics studies in 0.75PbMg1/3Nb2/3O3-0.25PbTiO3 showed good agreement with experimental data and revealed conflicts with the current polar nanoregion model. Here, we review our work and propose an alternate model for structure and dynamics in the relaxor phase.