Scientists and farmers in Uganda identified preferred sweet potato: (1) varieties through participatory varietal selection (PVS); and (2) new clones from seedling populations through a participatory plant breeding (PPB) approach. During these two processes, farmers identified 51 attributes of their landraces and of released varieties and used 21 criteria to select clones from amongst the seedling populations. Scientists had, in publications, listed attributes (11 main attributes identified), morphological descriptors (11) of released varieties and varietal needs (23) of sweet potato farmers. One released variety (NASPOT 1) was selected by farmers during PVS, mostly for its high and early yield of large, sweet and mealy roots, and several clones were selected through PPB amongst the seedling populations for a wider range of attributes. Some varietal attributes needed by farmers were not included by scientists either because they were very laborious, for example, selecting on-station for clones suitable for sequential piece-meal harvesting, or because occurrence of important abiotic or biotic stresses such as drought or pest damage were difficult to predict. Farmers seldom mentioned disease resistance but did mention pest resistance, consistent with easy visibility of both the causes of and the damage due to pests. Unlike scientists, farmers made no mention of a need for cultivars to have perceptually distinct features, despite this being a common attribute of landraces of most crops. © 2007 Springer Science+Business Media B.V.