Traditionally, scientists tend to approach cancer research in a reductionistic way: aiming at uncovering underlying, separate components in malignant processes. And indeed, great progress has been made by reducing the development of a tumor to single, specific genes and mutations. For instance, familial adenomatous polyposis (FAP) could be reduced to a germline mutation in the Adenomatous Polyposis Coli (APC) gene. The escape of tumor cells from immune surveillance could be reduced to the tumor expression of immune checkpoints, resulting in new approaches in tumor therapy by applying immune checkpoint inhibitors. However, a germline mutation in APC is not 1:1 related to colorectal cancer (CRC), and only some patients respond to immune checkpoint inhibitors. The point here is that biological systems, also comprising cancer, have properties that cannot be reduced to single components. The cooperation of the single components results in new, emergent properties. The outcome of an interaction in a complex network, like the immune system, depends on the many cell types involved and the numerous molecules that interact and activate or inhibit pathways. The way the composing elements are organized is a causal factor in itself for any emergent property. The rise of genomic analysis at the end of the previous century, enabling us to sequence a full genome at the DNA and RNA levels, has initiated an awareness of the need for ‘systems biology’: to consider a full system and how it is organized, in all of its aspects, to understand biological pathways and their outcomes. In this review, we outline the prospects and limitations of systems biology in cancer research and propose a causal framework that integrates upward and downward causation and multiple realizability to understand the emergent properties of tumors that determine the dynamics of tumor development.