This study examined empirically production effi ciency of cassava farms in Osun state of Nigeria using farm level data. The stochastic frontier production and cost function model was used to predict the farm level technical and economic effi ciencies respectively. The predicted technical effi ciency and economic effi ciency are the basis for estimating allocative effi ciency of the farms. Estimated results however, show that cassava farms in the study area exhibit decreasing positive return-to-scale giving the value of return to scale (RTS) of 0.840 obtained from the analysis, meaning that cassava farmers were effi cient in allocating their resources. Additionally, the analysis reveal that predicted effi ciency measure disaggregated into technical (TE), economic (EE) and allocative effi ciency (AE) with a view of examining not only TE but EE and AE when measuring productivity shows that mean TE, EE and AE of 0.903,0.89 and 0.807 were obtained from the analysis respectively meaning that TE appears to be more signifi cant than AE as a source of gain in EE. The policy implication of these fi ndings point to the fact that cassava farms in the study area were effi cient in allocating their resources considering their scope of operation and the limited resources.