Recursive elongation pathways produce compounds of increasing carbon-chain length with each iterative cycle. Of particular interest are 2-ketoacids derived from recursive elongation, which serve as precursors to a valuable class of advanced biofuels known as branched-chain higher alcohols (BCHAs). Protein engineering has been used to increase the number of iterative elongation cycles completed, yet specific production of longer-chain 2-ketoacids remains difficult to achieve. Here, we show that mitochondrial compartmentalization is an effective strategy to increase specificity of recursive pathways to favor longer-chain products. Using 2-ketoacid elongation as a proof of concept, we show that overexpression of the three elongation enzymes-, , and -in mitochondria of an isobutanol production strain results in a 2.3-fold increase in the isopentanol to isobutanol product ratio relative to overexpressing the same elongation enzymes in the cytosol, and a 31-fold increase relative to wild-type enzyme expression. Reducing the loss of intermediates allows us to further boost isopentanol production to 1.24 ± 0.06 g/L of isopentanol. In this strain, isopentanol accounts for 86% of the total BCHAs produced, while achieving the highest isopentanol titer reported for . Localizing the elongation enzymes in mitochondria  enables the development of strains in which isopentanol constitutes as much as 93% of BCHA production. This work establishes mitochondrial compartmentalization as a new approach to favor high titers and product specificities of larger products from recursive pathways.