Pasture improvement is a well-established technology to increase production in extensive livestock grazing industries by changing pasture composition and increasing soil fertility. The Cicerone Project farmlets located at Chiswick Research Station, near Armidale in New South Wales are providing valuable information at a credible scale on the response to three different management systems varying in levels of inputs and grazing management. The purpose of this paper is to outline a methodology for assessing the Cicerone Project and similar studies. The assessment focuses on stochastic efficiency of the different treatments. The impact of pasture persistence, climatic risk, and stochastic commodity prices on optimal rates of farm development are explored by using preliminary data from Cicerone to calibrate the GrassGro model. Essentially the farmlets modelled represent 2 technology packages. One is a moderate input package and the other is a high input package. Preliminary analysis indicates that direct comparison of the two farmlets may produce the wrong assessment, because one farmlet is operating at a suboptimal level of efficiency in a stochastic sense. This means that direct comparisons of technologies based on the field data may be biased since the technologies should be evaluated at the risk-efficient frontier. The concept of a risk efficient frontier is explained and applied to aid in identifying the trade-offs between profit and risk, and identify differences in the efficiency of the two farmlets.