Myrtle-leaf milkwort or sweet pea shrub (Polygala myrtifolia L.), family Polygalaceae, is a shrub from South Africa and is well adapted to Mediterranean-type conditions and used as an ornamental plant in gardens and pots or as cut flowers. During 2002 and 2003, mosaic symptoms and leaf distortion were observed in P. myrtifolia in Menton, Roquebrune-Cap Martin, Golfe Juan, and Antibes (Alpes Maritimes Department, France) in public gardens and potted plants. Occasionally, white streaks were observed in flowers. Cucumber mosaic virus (CMV) was identified in samples collected from the four locations on the basis of transmission to and symptoms exhibited by a range of diagnostic host plants (1), observation of isometric particles (≅30 nm) in crude sap preparations from the infected plants by electron microscopy, and positive reaction using double-antibody sandwich enzyme-linked immunosorbent assays (DAS-ELISA) with polyclonal antibodies raised against CMV (2). Each isolate was shown to be a group II CMV strain (3) using double-immunodiffusion analysis. During 2004, CMV was also detected using DAS-ELISA in P. myrtifolia samples collected in New Zealand (Christchurch, Akaroa, and Roturoa). To confirm that CMV was responsible for pathogenicity, the Menton isolate was isolated from local lesions on Vigna unguiculata, amplified in Nicotiana tabacum cv. Xanthi-nc, and then mechanically inoculated into 1-year-old P. myrtifolia, P. myrtifolia cv. Grandiflora, and P. myrtifolia cv. Compacta (synonymous to cv. Nana) plants. The D strain of CMV, a reference tomato strain from subgroup I (2), was used for comparison. All experimental plants were propagated from cuttings, grown hydroponically and all tested negative for CMV using DAS-ELISA prior to inoculation. At 12 weeks postinoculation, systemic symptoms were observed on leaves from all inoculated plants (10 plants per genotype for the Menton isolate and 5 plants per genotype for the D strain), except for two P. myrtifolia plants inoculated with the Menton isolate. CMV was detected in apical, noninoculated leaves using DAS-ELISA in all symptomatic plants. A total recovery from symptoms was observed in P. myrtifolia and P. myrtifolia cv. Grandiflora but not in P. myrtifolia cv. Compacta at 6 months postinoculation (mpi) in 7 of 15, 10 of 15, and 15 of 15 DAS-ELISA positive plants, respectively. At 7 mpi, the plants were pruned and planted in soil and at 8 mpi, CMV was detected using DAS-ELISA in most of the plants, and symptoms developed in a few stems of some of the plants. Tessitori et al. (4) described similar symptoms and have detected CMV in P. myrtifolia from Italy, but they did not reproduce the disease in healthy plants. Our results show that CMV is responsible for the symptoms observed and that both CMV subgroups are infectious in P. myrtifolia. Since P. myrtifolia is generally vegetatively propagated by cuttings, frequent CMV tests on the mother stock plants are recommended because of fluctuations in virus titer and symptom expression in some genotypes. To our knowledge, this is the first report of this CMV host in France and New Zealand. A voucher specimen will be deposited at the Station de Pathologie Végétale at INRA, Montfavet.