Many promising anti-cancer compounds fail in the clinical phase despite extensive testing in animal models. Reasons for this failure are varied, ranging from inadequate mouse models to dosing schemes that cannot be applied in humans. The apparent shortcomings of exisiting pre-clinical studies suggests a need for improved mouse models. Furthermore, the development of an increasing number of targeted drugs requires that biomarkers and/or the in vivo imaging of the tumor or its environment need to be established, ideally before the clinical candidate is selected. In this review, we focus on recent progress in mouse models, emphasizing imaging techniques, biomarker development, the tumor microenvironment and orthotopic metastatic tumor models. We discuss the application of the latter models to drug testing, making comparisons with classical subcutaneous xenograft models, and also models utilizing transgenic animals. Imaging modalities used to detect orthotopically implanted cells in the mouse, e.g. fluorescence or luciferase, which allow relatively high throughput measurements, will be compared to PET, CT, or MRI, which are closer to the clinical application. Examples for a combination of the latter imaging systems with luciferase or fluorescence in the mouse to develop or improve clinical imaging will be presented. Finally, the value of biomarkers in mouse models and strategies to identify novel targetspecific biomarkers in mouse models will be discussed.