Smallholder soybean production is investigated using an econometric analysis otherwise known as stochastic frontier analysis through transcendental logarithmic (translog) production function, which incorporates an inefficiency effects model. Ninety-six farmers were randomly selected through multistage techniques in Benue State, Nigeria. Factors (socio-economic and institutional) considered in the inefficiency effects model include household size, sex, age, years of schooling, farming experience in soybean production, health status, off-farm employment, non-family labour, credit accessibility, land fragmentation and extension contact. The parameters of the stochastic frontier translog production function are estimated contemporaneously with those involved in the inefficiency effects model. The results indicate that household size, age, non-family labour were significant and negatively related to the technical inefficiency while farming experience, off-farm employment, credit accessibility, land fragmentation, and extension contact were statistically significant and positively related to the inefficiency. The mean technical efficiency of the farmers is 0.84. This means that the farmers can still improve their efficiency level by 16%.