Density-specific sampling plans were developed under African conditions for the exotic spider mite, Mononychellus tanajoa (Bondar), a serious pest of cassava, Manihot esculenta. The within-plant distribution of Mononychellus tanajoa was found to favour new foliage, regardless of time of planting or plant age. Consequently, the first developed leaf near the top of the foliage was selected as the sampling unit and related to whole plant populations of M. tanajoa. The relationship between the mite population's variance and mean as measured by Taylor's Power Law proved to be stable over a range of planting dates, seasons and locations. Two binomial sampling plans, one based on Taylor's dispersion parameters and another based on direct field observations, were developed and compared. Binomial sampling, appropriate only for densities below 30 mites per leaf, was replaced by an enumerative procedure based on a ‘quick count’ protocol at higher mite densities.