In the present fundamental study, selectivity and efficiency of electromembrane extraction of 50 polar basic substances (-6.7 P < +1.0) was systematically studied for ten different supported liquid membranes. For each model substance, 23 molecular descriptors were collected and these were investigated as potential parameters for understanding of extraction efficiency and selectivity by means of partial least squares regression. Overall, a highly aromatic deep eutectic solvent composed of coumarin and thymol with addition of 2% ionic carrier (di(2-ethylhexyl) phosphate) provided the highest extraction efficiency with an average extraction yield of 69% from pure water samples, 55% from plasma, and 62% from urine. With this solvent system, ionic, cation-pi, and pi-pi interactions between the supported liquid membrane and analytes were dominant. Supported liquid membranes without aromaticity, however, operated primarily based on hydrogen-bonding interactions. This is the first time the relationship between analyte properties, solvent composition, and extraction yield has systematically been studied for polar bases in electromembrane extraction. This new knowledge represents a first step toward enabling future development and optimization of electromembrane extraction systems for polar bases based on rational design, rather than trial-and-error approaches.