The cell membrane plays a central role in the fitness and performance of microbial cell factories and therefore it is an attractive engineering target. The goal of this work is to develop a systematic framework for identifying membrane features for use as engineering targets. The metrics that describe the composition of the membrane can be visualized as “knobs” that modulate various “outcomes”, such as physical properties of the membrane and metabolic activity in the form of growth and productivity, with these relationships varying depending on the condition. We generated a set of strains with altered membrane lipid composition via expression of des, fabA and fabB and performed a rigorous characterization of these knobs and outcomes across several individual inhibitory conditions. Here, the knobs are the relative abundance of unsaturated lipids and lipids containing cyclic rings; the average lipid length, and the ratio of linear and non-linear lipids (L/nL ratio). The outcomes are membrane permeability, hydrophobicity, fluidity, and specific growth rate. This characterization identified significant correlations between knobs and outcomes that were specific to individual inhibitors, but also were significant across all tested conditions. For example, across all conditions, the L/nL ratio is positively correlated with the cell surface hydrophobicity, and the average lipid length is positively correlated with specific growth rate. A subsequent analysis of the data with the individual inhibitors identified pairs of lipid metrics and membrane properties that were predicted to impact cell growth in seven modeled scenarios with two or more inhibitors. The L/nL ratio and the membrane hydrophobicity were predicted to impact cell growth with the highest frequency. We experimentally validated this prediction in the combined condition of 42 °C, 2.5 mM furfural and 2% v/v ethanol in minimal media. Membrane hydrophobicity was confirmed to be a significant predictor of ethanol production. This work demonstrates that membrane physical properties can be used to predict the performance of biocatalysts in single and multiple inhibitory conditions, and possibly as an engineering target. In this manner, membrane properties can possibly be used as screening or selection metrics for library- or evolution-based strain engineering.