Diffusion-weighted magnetic resonance imaging (DW-MRI) has rapidly become an important part of cancer patient management. In this thesis, challenges in the analysis and interpretation ofDW-MRI data are investigated with focus on the intravoxel incoherent motion (IVIM) model, and its applications to childhood cancers. Using guidelines for validation of potential imaging biomarkers, technical and biological investigation of IVIM was undertaken using a combination of model simulations and in vivo data. To reduce the translational gap between the research and clinical use of IVIM, the model was implemented into an in-house built clinical decision support system. Technical validation was performed with assessment of accuracy, precision and bias of the estimated IVIM parameters. Best performance was achieved with a constrained IVIM fitting approach. The optimal use of b-values was dependent on the tissue characteristics and a compromise between bias and variability. Reliable data analysis was strongly dependent on the data quality and particularly the signal-to-noise ratio. IVIM perfusion fraction (j) was generally found to correlate with dynamic susceptibility contrast imaging derived cerebral blood volume. IVIM-f also presented as a potential diagnostic biomarker in discriminating between malignant retroperitoneal tumour types. Overall, the results encourage the use of IVIM parameters as potential imaging biomarkers.