The myotonic dystrophies are prototypic toxic RNA gain-of-function diseases. Myotonic dystrophy type 1 (DM1) and type 2 (DM2) are caused by different unstable, noncoding microsatellite repeat expansions - (CTG)DM1 in DMPK and (CCTG)DM2 in CNBP Although transcription of mutant repeats into (CUG)DM1 or (CCUG)DM2 appears to be necessary and sufficient to cause disease, their pathomechanisms remain incompletely understood. To study the mechanisms of (CCUG)DM2 toxicity and develop a convenient model for drug screening, we generated a transgenic DM2 model in the fruit fly Drosophila melanogaster with (CCUG)n repeats of variable length (n=16 and 106). Expression of noncoding (CCUG)106, but not (CCUG)16, in muscle and retinal cells led to the formation of ribonuclear foci and mis-splicing of genes implicated in DM pathology. Mis-splicing could be rescued by co-expression of human MBNL1, but not by CUGBP1 (CELF1) complementation. Flies with (CCUG)106 displayed strong disruption of external eye morphology and of the underlying retina. Furthermore, expression of (CCUG)106 in developing retinae caused a strong apoptotic response. Inhibition of apoptosis rescued the retinal disruption in (CCUG)106 flies. Finally, we tested two chemical compounds that have shown therapeutic potential in DM1 models. Whereas treatment of (CCUG)106 flies with pentamidine had no effect, treatment with a PKR inhibitor blocked both the formation of RNA foci and apoptosis in retinae of (CCUG)106 flies. Our data indicate that expression of expanded (CCUG)DM2 repeats is toxic, causing inappropriate cell death in affected fly eyes. Our Drosophila DM2 model might provide a convenient tool for in vivo drug screening.