The main purpose of preclinical study in non-human subjects such as animals is to predict effects for patient care. The predictive power of behavioral tests on drugs with brain effects is very low. Even within the same species, the replicability of many observations is questionable. These issues contribute to the limited progress in drug development for treating psychic disorders, such as depression, and physical brain injuries, such as stroke. An additional problem is that most behavioral tests impose varying degrees of ill-being on subjects. Combined with low test trustworthiness, this reduces public acceptance of animal research. Therefore, there is an urgent need for developing behavioral tests which assure maximal well-being for experimental subjects and offer improved predictive validity. This chapter will discuss these problems and present an alternative approach as a start for a solution: Incorporating seminatural environments in preclinical research will enhance external validity, hence generalizability between species, replicability, and animal welfare. Examples from the study of ketamine effects on rat sexual behavior and data on the importance of experimenter’s sex will be provided. This may facilitate new therapeutic developments and improve public standing of animal behavioral research.