Studies were conducted to examine distributional patterns ofBemisia tabaci(Gennadius) eggs and nymphs on two cultivars of upland cotton,Gossypium hirsutumL., and one American Pima cotton,G. barbadenseL., and to develop efficient sampling plans for estimating densities of immatures. On a per square centimeter basis, both eggs and nymphs were equally distributed among the four leaf sectors delineated by the major leaf veins. This pattern was independent of the nodal location of the leaf, cultivar, or sampling date. However, based on counts on 3.88-cm2disks near the petiole of the leaf, both egg and nymphal were aggregated at the proximal end of each sector. The relationship between disk and whole leaf counts varied with nodal position and cultivar. The greatest number of eggs and nymphs were found on mainstem leaves from nodes 2–4 and 4–7 (mainstem terminal = node 1), respectively. This pattern changed slightly with time but was similar among the three cultivars. The lowest coefficients of variation were associated with leaf counts from nodes 4–5 and 5–6 for eggs and nymphs, respectively. Based on variance partitioning and sampling cost analysis, a single 3.88-cm2disk from the base of the second sector of the fifth mainstern node leaf was determined to be the most efficient sample unit for estimating egg and nymphal densities. Sequential sampling stop lines were calculated for this sample unit using Green's (1970) method. Sample plan validation using Monte Carlo simulation indicated that actual levels of precision (SEM/mean) were poorer than those specified at low densities of immatures and better than specified at high densities. Further simulations indicated that stop lines for specified precisions of 0.20, 0.25, or 0.30 would maintain an average precision of 0.25 when egg or nymphal densities are100 per leaf disk, respectively. These sampling plans allow efficient monitoring for pest management application and will aid the study ofB. tabacipopulation dynamics in cotton.