Whole plant counts (visualin situexamination) for hooded beetles,Notoxus monodonF.; cotton fleahoppers,Pseudatomoscelis seriatus(Reuter); lady beetles,Hippodomiaspp.; and collops,Collopsspp., were made on ¼ acre of cotton to determine the best sampling-unit size. The insect counts were distributed according to the negative binomial, withkvalues increasing with sampling-unit size. The number of samples, linear row feet, and sampling time required for equal precision are illustrated for 7 sampling-unit sizes at 2 levels of precision. The single plant sampling-unit was determined as optimum for minimizing the variance.