This research presents a framework to describe and analyse risk and vulnerability to poverty of rural farm household in Northeastern of Thailand. Specific random sampling technique is used in the selection of 415 households, which are divided into 23 districts in Buri Ram province. It discusses the measurement of vulnerability to poverty, with an emphasis on quantitative techniques. The result on vulnerability

to poverty analysis is done by feasible generalized least squares (FGLS) method. Upon subjecting the data to analysis, the first stage of the OLS
reveals that 61.17% of the variation in log consumption (a measure of well-being) can be explained by the following factor: household size, household size square, aged dependency ratio, family member: bachelor education and above, number of children attend the school, unemployed household head, monetary assets, tangible asset value, ratio of rent of total expenditure, drought in the future, number of
last year risk occurred (2009), prolong sickness of household head (2009), chronic disease of other working family member (2009), flood (2009). The rest, 38.83%, can be attributed to the disturbance term. The results of the regression model by FGLS present the determinants of

vulnerability to poverty and variance of consumption. The comparison of the household consumption with Thailand poverty line in the year of 2009, that was at 1,586 Baht per capita per month. The rural headcount ratio in terms of household expected consumption less than poverty line is relatively high at 65.8%. The result indicates two groups of vulnerable households, which are, high and low vulnerable households. The estimates show that about 44.34% of households are vulnerable to poverty. The comparison of observed poverty status based on vulnerability index present that 75.2% of farm households are poor, whereas another 24.8% are non-poor.

Key word: Vulnerability, Risks, Farm Household, Feasible Generalized Least Square, Poverty.