||The principal advantage of in situ conservation is that it allows adaptive evolutionary processes to continue in the species that are being conserved. For a cultivated crop species, in situ conservation involves farmers' management of their own genetic resources even as the farmers themselves adapt to a changing environment. Improved seed selection practices and other on-farm breeding strategies have been proposed as a means of providing economic incentives for farmers to continue growing traditional varieties or landraces identified as important for conservation. this paper describes a pilot study among a group of indigenous farmers in the Sierra de Santa Marta, Veracruz, Mexico, who have collaborated in such efforts. The finding rise key issues about the potential impact of such an approach, as well as some useful methodological points for applied economists. In the study area, there is a high frecuency of experimentation, exchange, loss, and replacement of seed over ime - seed of the same varieties, including both modern and traditional varieties. This poses a challenge for economists models of varietal choice, wich tend to based on static perceptions of a "variety" as well as simplistic distinctions between "modern" and "traditional" varieties. Seed selection in the study area is not a single event but an iterative, continuous process. Women may be more involved in seed selection than previously though, wich may have implications for the welfare impact of new seed selection practices. Other implications of the study are that (1) the impact of introducing practices to enhance farmers' varieties is likely to be diffuse and difficult to observe, predict, or measure, and (2) in developing analytical models of farmer decision-making as it affects the diversity of genetic resources on the farm, the most apropriate unit of analysis for predicting the effects of some policy interventions is not likely to be the individual farmer or the individual farm household. A better understanding of the "social infrastructure" shaping seed and information flows is needed, since in the diffusion of innovations of this type, the seed system is based entirely on farmers and their interactions.