Sowing seed stocks with minimal virus content provides a key control measure in preventing damaging epidemics of Cucumber mosaic virus (CMV) in crops of narrow-leafed lupin (Lupinus angustifolius). A seed testing service provides an estimate of percent CMV infection based on a dry seed test in which bulked subsamples of ungerminated seed are ground to a fine powder for testing. When enzyme-linked immunosorbent assay (ELISA) was used, CMV antiserum that gave low background optical density (A(405)) values with extracts of powder from subsamples of healthy seed provided greatest accuracy, readily detecting one infected seed in subsamples of 100 seeds. In comparative ELISAs on duplicate subsamples from eight different seed stocks, germination and dry seed tests always gave similar percent infection values. When seed coats were separated from the embryos of CMV-infected and healthy lupin seeds before testing by ELISA, the virus was only detected in embryos from infected seeds and never in their seed coats. Treatment with trisodium phosphate did not alter the low ELISA optical density (A(405)) values obtained with seed coats separated from infected seeds. Therefore, seed coat contamination with CMV is lacking in lupin, justifying large-scale routine use of a dry seed test to estimate percent virus infection in commercial seed samples.