We have developed three approaches to the synthesis of proteins and protein-like macromolecules containing novel amino acids. In the first approach, we replace every copy of one of the natural amino acids by an analogue, in effect building proteins from an altered set of twenty starting materials. This approach is most useful when one is interested in changing the overall physical properties of the protein, or in de novo design of protein-based biomaterials. A second method, which has also been implemented successfully by Schultz and coworkers, allows site-specific incorporation of a single copy of an amino acid analogue in response to a stop codon. Such methods are useful in probing protein structure and function. The third approach, developed most recently, uses mutant transfer RNAs to break the degeneracy of the genetic code, and offers the prospect of a protein chemistry based on a substantially expanded set of amino acid building blocks. This lecture will describe the most important elements of each of these strategies as well as some thoughts on the design of wholly artificial proteins with potential application in biotechnology and materials science.