Agriculture sector in Ethiopia is characterized by its poor performance, while the population, which largely depends on agriculture, is growing at a faster rate. This necessitates looking for means to increase the productivity of smallholder farmers either by introducing new technologies or improving their technical efficiency at existing technology. And hence, this thesis provides new estimates of small holder farmers' technical efficiency and its principal determinants using a rural Tigray micro finance survey data collected in 2009. Both descriptive and econometric methods are used. The hypotheses tests confirm the adequacy of Cobb-Douglas over Translog frontier; the appropriateness of using SFA over OLS; the joint statistical significance of inefficiency effects; the appropriateness of using truncated normal distribution for one sided error; and the increasing returns to scale nature of the stochastic production function.