The potential of sweetpotato is limited by the perishable nature of the storage tubers. An extension of storage-life could be brought about, both by improving the storage environment and by breeding for extended shelf-life. The latter has the advantage in that it would result in an increased storage-life, both under conditions designed for long-term storage and under normal marketing conditions without incurring additional costs for producers and traders. To determine the feasibility of selecting cultivars for extended shelf-life, it is important to determine the genetic range in perishability among existing germplasm and the effect of the growth environment. In addition, a practical methodology for selection should be established. If possible, physiological characteristics associated with perishability should be identified to allow the development of indirect selection techniques. The progress of collaborative work conducted in Tanzania to address these issues is presented. Twenty-nine sweetpotato varieties, representative of the wide genetic variability of germplasm available in Tanzania, were assessed for their perishability. Under simulated Tanzanian marketing conditions, the major forms of deterioration observed were weight loss (primarily due to water loss) and rotting. Varieties varied considerably in both their rates of weight loss and of rotting, and these two characteristics were significantly correlated. Market observations have indicated that roots are subjected to considerable mechanical damage during normal transport and marketing. Thus, a range of varieties were tested for their rate of deterioration following simulated damage. In initial trials the ranking of varieties was not affected by the damage treatment, indicating that simulated damage need not be used during the selection procedure. These results suggest that indirect selection on the basis of physiological parameters may form part of the breeding programme in Tanzania.