Deletion of the INK4a/ARF locus at 9p21 is detected with high frequency in human melanoma. Within a short genomic distance, this locus encodes several proteins with established tumor-suppressor roles in a broad spectrum of cancer types. Several lines of evidence support the view that p16INK4a and p19ARF exert the tumor-suppressor activities of this locus, although their relative importance in specific cancer types such as melanoma has been less rigorously documented on the genetic level. Here, we exploit a well-defined mouse model of RAS-induced melanomas to examine the impact of germline p16INK4a or p19ARF nullizygosity on melanoma formation. We demonstrate that loss of either Ink4a/Arf product can cooperate with RAS activation to produce clinically indistinguishable melanomas. In line with the common phenotypic end point, we further show that RAS+ p16INK4a-/- melanomas sustain somatic inactivation of p19ARF-p53 and, correspondingly, that RAS+ p19ARF-/- melanomas experience high-frequency loss of p16INK4a. These genetic studies provide definitive proof that p16INK4a and p19ARF cooperate to suppress the development of melanoma in vivo.