||Descriptive data from the sample survey of households in the study communities enable us to profile the characteristics of: (1) farm households; (2) the maize populations they grow, based on their own descriptors and the perceptions of both men (production decision-makers) and women (consumption decision-makers);1 and (3) farmers’ seed selection and management practices. Farm households ♦ Households average between 4 and 6 members, and the mean age for production decisionmakers is about 50 years. All production and consumption decision-makers are Spanishspeaking. Indigenous languages are used more in Mazaltepec and Santa Ana. ♦ By some indicators of wealth (television, refrigerator, electricity, gas stoves), from a global perspective, survey households are not poor. ♦ Huitzo depends more on local nonfarm income than other survey sites; Huitzo and Valdeflores depend less on local agricultural production; Huitzo and Mazaltepec depend less on remittances. In San Lorenzo, Santa Ana, Valdeflores, and Amatengo, at east one-fourth of households reported that remittances are an important source of income. ♦ All sites except Huitzo reported average per capita maize requirements higher than the national mean. The average farm household in the survey communities is a net seller in some years and a net consumer in others. The market for maize appears to be local. ♦ The average farm size in 1996 was 3.5 ha with 3 ha of maize. The mean number of soil types per farm, as well as the extent of land fragmentation, was lower in Huitzo and Amatengo. Fragmentation seems to be highest in Santa Ana and Valdeflores. Tenure arrangements contrast sharply between communities.Maize populations ♦ Maize populations grown by farm households have been classified based on farmer taxonomy into 5 classes of Blanco (white-grained) maize, 3 classes of Amarillo (yellow-grained) maize, 1 Negro (black or purple-grained) maize, 1 Belatove (pinkish-grained) maize, 1 Pinto (grain of mixed color), 1 Mejorado (improved) class and a Tepecentle variety. Tepec ntle is a distinct maize race, and all other classes are found within the Bolita racial complex. ♦ Blanco types occupy over 80% of the area and representing two-thirds of the seed lots planted in 1997. Improved maize was grown almost exclusively in Huitzo, which has the most irrigated land. Even in Huitzo, it occupied only 7% of the area in the survey year. Subjective yield distributions suggest that: (1) improved maize dominates local types; (2) Blanco types dominate colored-grain types; and (3) consumption partners (women) are significantly more pessimistic than production partners (men). The yield distributions of Santa Ana are significantly different from those of Huitzo, Mazaltepec, and Amatengo. Farmers in the survey communities grow maize primarily for food or feed rather than grain sales, and they are interested in many characteristics in addition to yield. While they rate improved maize well in terms of grain yield and fodder, they rank it as a poor supplier relative to local varieties. Among the ocal varieties, Blanco types were rated superior to colored-grain types with respect to grain yield per hectare, suitability for sale and most consumption characteristics. Amarillo was highly rated for tlayudas, feed, and fodder, and Negro and Belatove have shorter growing seasons. Although men and women rank the importance of characteristics differently, four of the top five characteristics are the same for both sexes: (1) drought tolerance; (2) resistance to insects in storage; (3) produces “something” even in bad years; and (4) grain weight. Men added grain yield per hectare and women added the taste of tortillas to the set of most important characteristics. Seed management Farmers know their varieties—they have grown them for an average of over 20 years. The concept of “own seed” is ambiguous, however, since a large proportion of farmers also combine and replace seed lots (see definition of terms in Methods section). The highest propensity to give, exchange, combine, or replace seed was found in Sant Ana, although these practices were also observed in San Lorenzo and Amatengo. Exchange is primarily local. As has been found elsewhere in Mexico, the seed selection criteria used by farmers are those related to grain and ear health, grain size, grain filling, and ear size. Less than half of survey farmers reported that they separated food or feed grain from seed at harvest time. The most frequent form of selection is the continual separation of good ears from those removed every few days for preparation of nixtamal. Perceptions differ between men and women regarding responsibilities for seed management and selection, but women’s role is likely to be substantial in this separation activity. There is evidence that farmers are exerting strong indirect selection pressure for resistance to insect damage in storage, but no direct pressure on husk cover. Husk cover is important as a “first-line defense.” Maize diversity Farm-level diversity appears to be greatest in Santa Ana and San Lorenzo and least in Amat ngo and Valdeflores, as measured by numbers of varieties per farm and by a Simpson index based on area shares. Community-level diversity bears no direct relationship to farm-level diversity because of differences in the scale of measurement. Diversity remains high at the community level in San Lorenzo and is relatively low in Santa Ana. There is no strong evidence that farmers in these communities recognize a loss of maize populations during the past two decades.