The objective of this study was to evaluate the toxicity of pre-emergent herbicides to eucalyptus plants and weed control efficiency when applied via irrigation water in crop implementation. The experiment was conducted under field conditions, using the experimental randomized block design for each herbicide, in a 5 x 2 bifactorial scheme, with four replications, considering the five evaluated herbicide doses as factors (isoxaflutole – 0; 75; 105; 135 and 165 g a.i. ha-1; flumioxazin – 0; 70; 90; 110 and 130 g a.i. ha-1; sulfentrazone – 0; 400; 600; 800 and 1000 g a.i. ha-1), besides two forms of application (directly on the plant and only on the soil). The treatments were applied manually 20 days after the seedlings were planted, using a watering can. Eucalyptus intoxication percentage, plant height, stem diameter, Falker chlorophyll index, leaf number, leaf area and shoot dry mass were evaluated. Weeds present in the useful area around the planting holes were identified and quantified to determine the plant density in the area sampled per treatment. It was found that only sulfentrazone caused plant intoxication symptoms, despite not influencing eucalyptus development. Isoxaflutole, regardless of application form, and flumioxazin, applied directly on the plants, were selective to the crop, providing at the highest doses (165 and 130 a.i. ha-1, respectively) efficient weed control and increment in initial eucalyptus growth.