Studying the sources of growth in agricultural production, examining the extent of inefficiency, and identifying the sources of such inefficiency, is an important step forward to improve the livelihood of subsistence farm households in developing countries. A stochastic frontier analysis is used because, in addition to accounting for sources of growth in agricultural output, this method explicitly incorporates efficiency differences in the analysis. The empirical analysis uses panel data from the Ethiopian Rural Household Survey collected during 1994 through 2009. The results indicate that most of the increase in agricultural output is attained by increased use of traditional inputs such as size and quality of cultivated land, labor, numbers of oxen and hoes, and was heavily influenced by amount of precipitation received. By contrast, the rate of fertilizer application contributed little for increase in output. Participation in the extension program made moderate contribution towards increases in output. Each agroecological zone included in the study gained from Hicks-neutral technological improvements during the 1994–2004 period. Nonetheless productivity levels in 2009 were not different from levels in 1994, and they had declined between 2004 and 2009. Average level of farming efficiency for the surveyed farmers across all the years was 0.46, indicating that an average farmer produces less than half of the value of output produced by the most efficient farmer using the same technology and inputs. However, average farming efficiency has improved during the 1995–2009 period. Farm households’ level of farming efficiency is improved by reducing labor bottlenecks and increased education. Households that have diversified risk from plots that are located sufficiently apart appear more efficient. Households that own more animals both in terms of two or more ploughing oxen or total livestock units are more efficient. Drought affects efficiency adversely whenever it strikes. Farmers that live in close proximity to markets are less efficient. On average, farming inefficiency has consistently declined in the period considered. The results suggest that each agroecological zone is faced with different opportunities and obstacles.