Using stochastic frontier analysis, efficiency of production of wheat in Western Australia was studied. The production function model used was a relatively simple input model, consisting of wheat yield, effective rainfall, fertilizer application rates and year of study. Inefficiency was captured in a second model that incorporated machinery capital investment, opening equity level, and year of study. Data covered the production years 2004 through to 2007. The results demonstrated that inefficiency was present in wheat production in Western Australia and that inefficiency increased over the period from 18% in 2004 to 29% in 2007. Higher machinery investment per hectare and opening equity levels reduced inefficiency, due to producers having sufficient capacity, mechanical or financial, to adapt to variability within the production season. The results demonstrated the stochastic nature of efficiency and that for some firms improving efficiency may not be possible or feasible due to limitations within the firm. This also holds for firms that are relatively efficient in some years and that the reasons for the inefficiency are not necessarily production related, hence, programs targeted to improve efficiency may not be very successful. On the other hand firms that are consistently inefficient provide an ideal target audience for programs to improve efficiency. However, these programs must be conditioned on adequately identifying the source(s) of inefficiency and the producer having access to resources to increase efficiency. Similar analyses could be undertaken in different crops or different geographic locations, to identify if and why inefficiencies are present in other production systems.