Often in science, what seem to be definitive answers lead to new questions, which lead to new answers and the cycle goes on. That's what happened when I came across heat shock proteins. I was a postdoc at Sloan Kettering in 1985, and I had just received the results of the N-terminal sequence of gp96, a tumor-rejection antigen, which I had purified from a mouse sarcoma, and earlier from a rat hepatoma. With the sequence in hand, I felt confident that I had just solved the longest standing puzzle in cancer immunology – the structural basis of the immunogenicity of methylcholanthrene-induced cancers[1] – I banged out this sentence on the nearest typewriter (see photo) and pasted it into my lab notebook. Little did I know that the journey had just begun. What I didn't know then, but would later find out, was that the immunogenicity didn't derive from gp96 but from the peptides chaperoned by it, and that this is a general property of this whole class of proteins. Twenty years later, I remain mesmerized by these molecules as their remarkable properties continue to unfold.