In its first production phase, The ENCODE Project Consortium (ENCODE) has generated thousands of genome-scale data sets, resulting in a genomic "parts list" that encompasses transcripts, sites of transcription factor binding, and other functional features that now number in the millions of distinct elements. These data are reshaping many long-held beliefs concerning the information content of the human and other complex genomes, including the very definition of the gene. Here I discuss and place in context many of the leading findings of ENCODE, as well as trends that are shaping the generation and interpretation of ENCODE data. Finally, I consider prospects for the future, including maximizing the accuracy, completeness, and utility of ENCODE data for the community.