The interference imposed by weeds can cause damages to cassava development. Thus, adopting control measures is fundamental. The use of pre-emergence herbicides is one of the control alternatives, since cassava presents a good tolerance to herbicides applied in this modality. However, herbicides may present differential selectivity due to the variety of cassava that is planted. The objective of this study was to evaluate both weed control effectiveness and selectivity of different herbicide combinations applied at pre-emergence of two cassava varieties. Two experiments were established in the field using a randomized complete block design, in a split plot scheme, with four replications. In the efficacy experiment, application of six combinations of herbicides, including clomazone + ametryn, clomazone + metribuzin, clomazone + flumioxazin, isoxaflutole + ametryn, isoxaflutole + metribuzin, isoxaflutole + flumioxazin were evaluated in the main plot. In addition, a non-treated control and a weeded control were evaluated. In the subplots, the Caravela and Pretinha varieties were planted. For the selectivity experiment, a non-treated control was not included. The combinations containing clomazone presented a longer residual period of weed control as well as a better control of benghal dayflower. Metribuzin exhibited lower efficacy than other herbicide combinations containing clomazone or isoxaflutole. The Pretinha variety showed greater tolerance to weed interference. Every herbicide combination that was applied in cassava pre-emergence presented selectivity for both Caravela and Pretinha varieties.