The Red Queen hypothesis depicts evolution as the continual struggle to adapt. The hypothesis, commonly invoked to explain organismal evolution, can also be applied to genic evolution. According to this hypothesis, new genes, especially those originating from non-genic sequences (i.e., de novo genes), would be eliminated unless they evolve continually to adapt to a changing world. Non-coding genes, represented by microRNAs (miRNAs), are the most common de novo genes. Here, we analyze six Drosophila de novo miRNAs that are testis-specific. These miRNAs exhibit a very high rate of evolution in their DNA sequence, transcript production, and expression pattern. We knock out two of the youngest miRNAs and observe three patterns. 1) The fitness advantage of miR-984 presence seems to have vanished in D. melanogaster even though its DNA sequence bears a signature of past adaptation. 2) Another gene, miR-983, appears to have become maladaptive in D. melanogaster as the fitness of the knockout mutant increases. 3) In D. simulans, the deletion of miR-983 is associated with extensive mis-regulation of male reproductive genes, resulting in a 70% loss in male fertility. The fitness contributions of these genes are, respectively, neutral, negative, and positive. As predicted by the Red Queen hypothesis, de novo genes either evolve rapidly or face elimination.