When selecting a weed chemical treatment (herbicide, product mixtures, dose, or application timing), an important issue to consider is its selectivity to the crop of commercial interest. The objective of this study was to evaluate the selectivity of the herbicide metribuzin to carrot plants as a function of genotype, dose, and plant growth stage at the time of application. Two experiments were carried out, one in a greenhouse and another in the field. The greenhouse experiment was arranged in a completely randomized, 5 x 2 x 3 factorial design, with four replications. Metribuzin doses (0, 72, 96, 144, and 192 g ha-1) were sprayed on plants with 2-3 and 4-5 true leaves of the cultivars Maestro, BRS Planalto, and Verano. Cultivar Maestro was cultivated in the field in a 2 x 5 factorial experiment in randomized complete block design, with four replications. Carrot plants, at two stages of development (2 and 5 true leaves), were sprayed with the same doses of metribuzin applied in the greenhouse experiment. Metribuzin, regardless of application time and dose tested, was selective for cultivars BRS Planalto and Verano (greenhouse) and Maestro (both greenhouse and field), without reduction in quality and yield of roots.