All living organisms share similar reactions within their central metabolism to provide precursors for all essential building blocks and reducing power. To identify whether alternative metabolic routes of glycolysis can operate in E. coli, we complementarily employed in silico design, rational engineering, and adaptive laboratory evolution. First, we used a genome-scale model and identified two potential pathways within the metabolic network of this organism replacing canonical Embden-Meyerhof-Parnas (EMP) glycolysis to convert phosphosugars into organic acids. One of these glycolytic routes proceeds via methylglyoxal, the other via serine biosynthesis and degradation. Then, we implemented both pathways in E. coli strains harboring defective EMP glycolysis. Surprisingly, the pathway via methylglyoxal immediately operated in a triosephosphate isomerase deletion strain cultivated on glycerol. By contrast, in a phosphoglycerate kinase deletion strain, the overexpression of methylglyoxal synthase was necessary for implementing a functional methylglyoxal pathway. Furthermore, we engineered the ‘serine shunt’ which converts 3-phosphoglycerate via serine biosynthesis and degradation to pyruvate, bypassing an enolase deletion. Finally, to explore which of these alternatives would emerge by natural selection we performed an adaptive laboratory evolution study using an enolase deletion strain. The evolved mutants were shown to use the serine shunt. Our study reveals the flexible repurposing of metabolic pathways to create new metabolite links and rewire central metabolism.