Farmers in the high plateau region of Algeria are assumed to exhibit risk averse behavior, particularly, due to highly variable weather conditions inducing income instability over time. This in turn directly affects their production behavior. The Eastern High Plateau (Setif) is not a homogeneous region. In the El-Eulma daira, for example, three different agroecological zones have been identified on the basis of climate, topography and soil quality. In addition, two distinct agricultural sectors (private and socialist) coexist side by side in each of the agroecological zones. This study constitutes an attempt to measure farmers' risk attitudes in three communes (El-Eulma, Oum Ladjoul and Beni Fouda) which are representative of the three agroecological zones of the El-Eulma daira. Farmers' risk attitudes were measured through the experimental approach developed by Binswanger in India. The technique used consisted of presenting the subjects, i.e. the farmers, with a set of alternative prospects involving real money. Based on the derived risk aversion coefficients, a series of tests was run to determine if farmers' risk attitudes are dependent on the zone and/or the sector. The effect of socioeconomic characteristics (age, schooling, number of working children, etc.) on partial risk aversion was analyzed. Finally, the derived risk aversion coefficients were used in a risk programming model (MOTAD) to determine optimal farming plans for private as well as socialist sector farmers. The experiment results indicate that regardless of the zone and the sector, farmers unanimously exhibit risk averse attitudes. At low payoff level, the distribution of risk preferences is more spread. A narrower distribution occurs at higher payoff levels (e.g. 200 DA scale). There was no evidence of significant difference among sites and between sectors. Also socioeconomic attributes correlate poorly with the estimated partial risk aversion coefficients. In the socialist sector major discrepancies between the risk programming model solutions and actual activity levels occured. They were expected because of the specific structure of this sector. The inclusion of government cropping pattern recommendations in the constraint matrix indicates that government interventions have a different effect on socialist farmers' welfare of the three zones. / Graduation date: 1986