Cassava (Manihot esculenta Crantz) is a carbohydrate staple and cash crop for about 800 million people in the tropics. As food, its use is influenced by its content of potentially toxic cyanogenic glycosides that release hydrocyanic acid when enzymatically hydrolysed (Conn, 1969). Cyanide plays important role in the protection of cassava plants from attacks by animals and insect pests. The glycosides are synthesized mainly in leaves and translocated to all parts of the cassava plant (Bediako, Tapper & Pritchard, 1981). De Bruijn (1973) noted more than a 100% increase in the HCN content of stem bark above the incision after ringing, especially during the first 2 days, and a continued increase for at least 2 months. However, when leaves were removed, no such increase was observed. De Bruijn (1973) observed that younger plants showed a greater increase (165%) than older plants (65%). Leaf HCN did not increase after ringing, whereas root HCN decreased by about 20% in 2 weeks. The literature on cyanogenic glycoside translocation is relatively recent and limited. Consequently, the exploitation of the supposed direct relationship between leaf HCN and root HCN in the selection of low HCN clones in cassava breeding programmes needs to be carefully assessed. The assessment is complicated because the methods used by different workers for the determination of HCN vary in their efficiency. Thus, Cooke, Howland & Hahn (1978) found no correlation (r2 = 0·13) between leaf HCN and root HCN among 108 clones, yet over 88000 genotypes have been screened by analysis of leaf HCN, and leaf analysis has been used routinely in cassava improvement for 9 years (Hahn, 1983). Hahn also reported the range of HCN concentration within each leaf-picrate class (1 = 80, 2 = 80−200, 3 = 200 mg/100g fresh weight) but the ranges within each of these classes were too wide to enable effective selection for low HCN clones.