In metabolic modeling, experimental data is used to constrain the solution space of allowable flux distributions to biologically feasible phenotypes. This makes metabolic modeling a powerful tool for integrating information from different experimental sources. In this study we integrate transcriptomics with metabolomics for the first time. For this, we generated a new formulation of transcriptomics-based constraints (derived from IMAT), and we integrated Thermodynamics-Based Metabolic Flux Analysis (TMFA) constraints into an E. coli Genome Scale model. We found that when used separately, thermodynamic and transcriptomic constraints produced contradicting results. The optimal solutions for a model with transcript constraints alone were not thermodynamically feasible. This conflict forces us to choose between a solution space that is “optimally” constrained by transcriptomics or one that is thermodynamically feasible. All models should be thermodynamically consistent, whereas the integration of transcriptomics is only desirable. Adding transcription constraints is complicated because of the large amount of thresholds and/or assumptions that are required by the currently available methods. The proposed method enforces TMFA and transcript-based constraints together, ensuring solutions that are feasible for both data sets simultaneously. The combination reduces the solution space much more than each method would individually. Hence, we can reduce the bias introduced by transcriptomics-based constraints, while still obtaining a desired level of flexibility in our model.