Cyanogenesis, the release of hydrogen cyanide from damaged plant tissues, involves the enzymatic degradation of amino acid-derived cyanogenic glucosides (alpha-hydroxynitrile glucosides) by specific beta-glucosidases. Release of cyanide functions as a defense mechanism against generalist herbivores. We developed a high-throughput screening method and used it to identify cyanogenesis deficient (cyd) mutants in the model legume Lotus japonicus. Mutants in both biosynthesis and catabolism of cyanogenic glucosides were isolated and classified following metabolic profiling of cyanogenic glucoside content. L. japonicus produces two cyanogenic glucosides: linamarin (derived from Val) and lotaustralin (derived from Ile). Their biosynthesis may involve the same set of enzymes for both amino acid precursors. However, in one class of mutants, accumulation of lotaustralin and linamarin was uncoupled. Catabolic mutants could be placed in two complementation groups, one of which, cyd2, encoded the beta-glucosidase BGD2. Despite the identification of nine independent cyd2 alleles, no mutants involving the gene encoding a closely related beta-glucosidase, BGD4, were identified. This indicated that BGD4 plays no role in cyanogenesis in L. japonicus in vivo. Biochemical analysis confirmed that BGD4 cannot hydrolyze linamarin or lotaustralin and in L. japonicus is specific for breakdown of related hydroxynitrile glucosides, such as rhodiocyanoside A. By contrast, BGD2 can hydrolyze both cyanogenic glucosides and rhodiocyanosides. Our genetic analysis demonstrated specificity in the catabolic pathways for hydroxynitrile glucosides and implied specificity in their biosynthetic pathways as well. In addition, it has provided important tools for elucidating and potentially modifying cyanogenesis pathways in plants.