The methyl-accepting chemotaxis proteins are a family of receptors in bacteria that mediate chemotaxis to diverse signals. To explore the plasticity of these proteins, we have developed a simple method for selecting cells that swim to target attractants. The procedure is based on establishing a diffusive gradient in semi-soft agar plates and does not require that the attractant be metabolized or degraded. We have applied this method to select for variants of the Escherichia coli aspartate receptor, Tar, that have a new or improved response to different amino acids. We found that Tar can be readily mutated to respond to new chemical signals. However, the overall change in specificity depended on the target compound. A Tar variant that could detect cysteic acid still showed a strong sensitivity to aspartate, indicating that the new receptor had a broadened specificity relative to wild-type Tar. Tar variants that responded to phenylalanine or N-methyl aspartate, or that had an increased sensitivity to glutamate showed a strong decrease in their response to aspartate. In at least some of the cases, the maximal level of sensitivity that was obtained could not be attributed solely to substitutions within the binding pocket. The new tar alleles and the techniques described here provide a new approach for exploring the relationship between ligand binding and signal transduction by chemoreceptors and for engineering new receptors for applications in biotechnology.