Malignant cells characteristically exhibit altered metabolic patterns when compared with normal mammalian cells with increased reliance on anaerobic metabolism of glucose to lactic acid even in the presence of abundant oxygen. The inefficiency of the anaerobic pathway is compensated by increased glucose flux, a phenomenon first noted by Otto Warburg approximately 80 years ago and currently exploited for 2-fluoro-2-deoxy-D-glucose-positron emission tomography imaging in clinical radiology. The latter has demonstrated the glycolytic phenotype is a near-universal phenomenon in human cancers. The potential role of the glycolytic phenotype in facilitating tumor invasion has been investigated through mathematical models of the tumor-host interface. Modified cellular automaton and diffusion reaction models demonstrate protons will diffuse from the tumor into peritumoral normal tissue subjecting nontransformed cells adjacent to the tumor edge to an extracellular pH significantly lower than normal. This leads to normal cell death via p53-dependent apoptosis pathways, as well as degradation of the interstitial matrix, loss of intercellular gap junctions, enhanced angiogenesis, and inhibition of the host immune response to tumor antigens. Transformed cells maintain their proliferative capacity in acidic extracellular pH because of mutations in p53 or some other component in the apoptosis pathways. This allows tumor cells to remain proliferative and migrate into the peritumoral normal tissue producing the invasive phenotype. Mathematical models of invasive cancer based on tumor-induced acidification are consistent with extant data on tumor microenvironment and results from clinical positron emission tomography imaging, including the observed correlation between tumor invasiveness and glucose utilization. Novel treatment approaches focused on perturbation of the tumor microenvironment are predicted from the mathematical models and are supported by recent clinical data demonstrating the benefits of azotemia and metabolic acidosis in survival of patients with metastatic renal cancer. The evolutionary basis for adoption of the glycolytic phenotype during carcinogenesis remains unclear because it appears to confer significant competitive disadvantages on the tumor cells due to of inefficient energy production and expenditure of resources to remove the acid byproducts. We propose that the glycolytic phenotype represents a successful adaptation to environmental selection parameters because it confers the ability to invade. That is, the glycolytic phenotype allows the cell to move from the microenvironment of a premalignant lesion to adjacent normal tissue. There it competes with normal cells that are less fit than the populations within the tumor in a microenvironment of relative substrate abundance. The consequent unrestrained proliferation allows the glycolytic phenotype to emerge simultaneous with the transition from a premalignant lesion to an invasive cancer.