Using three rounds of data from the Ethiopia Socioeconomic Survey (ESS), we estimate a series of dynamic random effect probit models accounting for unobserved heterogeneity and the initial conditions problem using four measures of food insecurity and poverty. The descriptive analysis suggest higher levels of downward mobility and lower rates of exit for the food insecurity measures relative to food poverty and general poverty. There also appear to be lower levels of persistence in the subjective measure of food insecurity although for all measures close to the majority of the sample experience at least one period of food insecurity, food or general poverty. The estimation results provide some evidence of persistence or state dependence exists for subjective food insecurity, food and relative poverty measures. There is also evidence of cross effects so that being subjectively food insecure in the past increases the probability of being food poor and relatively poor in the current period. Keywords: Food insecurity; Poverty; State dependence; Ethiopia; farm households JEL codes I32, Q12, C23