Since 10 years ago, when the seven hallmarks of cancer were first defined by Hanahan and Weinberg, after decades of molecular, cellular and clinical investigations, new systems-based approaches have provided a wide range of improved investigative methods. These approaches integrate various global data types into mathematical and computational models of molecular and cellular pathways and networks that become dysregulated in cancer, since the models are now able to take into account the large-scale properties of complex biological networks. Genome variation and instability have been revisited through study of genetic and genomic networks; while transcription and protein interaction networks are revealing cancer biomarkers of modular change. Growth, proliferation and apoptosis are being more fully described by signalling network modelling. Sustained angiogenesis and metastasis are being addressed via multiscale modelling. Enhanced understanding of the initial hallmarks of cancer, extended to the control of metabolism and stress, is opening novel avenues for cancer diagnosis and treatment. It is fully expected that further progress will take place through iterative cycles of experimentation and modelling, typical of systems biology. All of this will require advances in molecular data acquisition, multiscale integration of data scales and types, new approaches to data analysis and improved modelling. Success in all these endeavours cannot be achieved without better cross-disciplinary interactions among researchers and technologists.