This study examines the association of livelihood dynamics of rural households in Ethiopia with rainfall conditions and socio-economic characteristics, using a 15-year panel data set. We employed fixed and random-effect conditional logit models to explain household decision-making processes regarding livelihood strategies. Our finding shows that participation of rural households in non-farm livelihoods has been increasing over the years but with great fluctuations. We also found that rainfall conditions during the main rainy season negatively and significantly affect household decisions to pursue non-farm livelihoods. The motivation of farm households to diversify into non-farm livelihoods is mainly driven by low-performance farming outcomes as well as demographic characteristics (specifically adult household size, human capital and education) and degree of access to financial schemes. These findings suggest policy implications for increasing access to financial schemes and improving household-member skills through vocational training and education to enable them to engage in high-return and profitable non-farm livelihoods. Acknowledgement :