Using panel data from the Mexican Family Life Survey, this paper estimates a multinomial logistic regression model to analyze the dynamics of chronic and transient poverty in Mexico. Based on the spells approach, transition matrices are constructed to observe households’ entry into and exit from poverty and multinomial logistic regression is used to analyze which factors explain the dynamics of poverty in Mexico. It was found that 36% of households are chronically poor and 64% are transiently poor. Also, we found that the variables directly related to chronic poverty are: belonging to an ethnic group, living in a rural area, a large family size, having a high percentage of older adults and children in the household and having a female household head. On the other hand, it was found that having more education, the age of the household head and having access to potable water and electricity in the household are positively related with the probability of escaping poverty.