This paper summarizes a study of the application of various two-stage sampling designs and estimation procedures for preparation of state estimates of agricultural items in North Carolina. Among the principal objectives of the investigation were (i) the comparison of selection of the primary sampling units with equal probability and with probability proportional to size and (ii) assessment of the magnitude of the within primary sampling unit error component contribution to the total survey sampling error. Altogether eight different survey designs with associated estimates have been used to secure information on these two main points. Examination of these designs and estimates indicates a preference for a particular ratio estimate.