One mechanism for the origin of new plant microRNAs (miRNAs) is from inverted duplications of transcribed genes. However, even though many young MIRNA genes have recently been identified in Arabidopsis thaliana, only a subset shows evidence for having evolved by this route. We propose that the hundreds of thousands of partially self-complementary foldback sequences found in a typical plant genome provide an alternative path for miRNA evolution. Our genome-wide analyses of young MIRNA genes suggest that some arose from DNA that either has self-complementarity by chance or that represents a highly eroded inverted duplication. These observations are compatible with the idea that, following capture of transcriptional regulatory sequences, random foldbacks can occasionally spawn new miRNAs. Subsequent stabilization through coevolution with initially fortuitous targets may lead to fixation of a small subset of these proto-miRNA genes.