Potato production from true seed, or true potato seed (TPS) is not a new system for this vegetatively propagated crop. Andean farmers have grown potatoes in the cool highlands from true seed for many centuries. TPS appears today as a promising technology in the warm tropics, where diseases impede the production of pathogen-free potato propagules. However, the identification of suitable genetic resources for potato seed production requires the selection of clones that lack antagonism between tuber growth and seed embryo development. Furthermore, the potential TPS parents need to possess characteristics of a sexually propagated crop that is grown for the harvest of its vegetative part, i.e., the tuber. This new process of 'domestication' requires a change in the selection methods, which starts by assessing the potential of available potato clones for both reproductive characteristics and agronomic traits, particularly those of the tuber. Tetraploid landraces and cultivars of the pathogen-tested collection of the Centro Internacional de la Papa were assessed for reproductive and agronomic characteristics during three consecutive years in San Ramón, a Peruvian location in the warm lowland tropics. The clones showed significant variation for all characteristics recorded. The Shannon-Weaver diversity index (H) was highest for flowering duration (0.617) and intensity (0.600). The lowest H was calculated for bee attraction in this location. The overall H was not significantly affected by the year-to-year variation, but some of the characteristics recorded (e.g. flowers per inflorescence, anther type, style length, plus fruit and berry set) showed distinct H according to the year of assessment. Some of the cultivars with promising TPS traits, as determined by their reproductive characteristics, also had outstanding tuber set, weight and yield. Furthermore, hybrid offspring from the selected clones from the pathogen-tested elite sample, did show high tuber yield with appropriate tuber set and reproductive traits as shown by progeny testing. Hence, the approach followed for identifying parental sources as discussed in this article appears to be appropriate.