This thesis introduces a framework for building modular mRNA logic devices that can be used in synthetic biological systems. Such devices require two primary components: a sensor domain and actuator domain. The sensor can be adapted from an aptamer, which is an RNA structure that binds a particular ligand. These aptamers can be selected through in vitro techniques. Here we study the effects of both magnesium (important for in vivo operation) and the ligand character itself on the binding affinity and probability of finding a high affinity aptarner. Understanding these effects will allow us to select which parts of metabolic pathways are most amenable to selecting good sensors. We find that molecular targets with high molecular weight, and those with less degrees of freedom (rotatable bonds) are easier to bind that small, flexible molecules. Further, we observe that aptamers selected to bind with high affinity also bind well at physiological concentrations of magnesium. The informational complexity needed to bind at physiological magnesium is small relative to the informational complexity required to bind small ligands relative to larger ligands. The actuator domain is an RNA enzyme that transduces the binding event of the sensor domain into a change in translation. By locating the sensor physically on the transcript, control will be directly coupled to the output. An aptamer selected here to bind p-amino-phenylalanine (pAF) was coupled via a random bridge sequence to an S. mansoni ribozyme and selected for sequences which cleave in the presence of pAF. Several sequences were isolated, which exhibit up to a 1.5-fold difference in fluorescence when expressed in vivo to regulate dsRedExpress. The specificity of the sensor domain allows for scalability in the number of different devices capable of operating independently within a cell. Both of these features contribute to the overall goal of building complex, synthetic biological systems.