It has been recognized that systemic and local infections of the reproductive tract, as well as the inadvertent introduction of microorganisms during processing, may potentially contribute to the contamination of semen (Bielanski 2007 Theriogenology 68, 1-22). A number of viral pathogens have been identified in association with semen, such as bovine viral diarrhea virus (BVDV). The aim of the study was evaluate the capacity of sperm separation procedures Percoll gradient and swim-up, as proposed by Bielanski et al. (1992 Reprod. Domest. Anim. 27, 303-306) and also a combination of swim-up and Percoll, to reduce or eliminate BVDV from experimentally infected semen samples. We used frozen semen straws (0.5 mL) and the virus sample was cytopathic type BVDV (106.68 TCID50mL-1). The experimental infection was performed immediately after semen thawing in a 1 : 1 volume proportion semen/viral suspension. The experimental groups were Percoll control (Pc); swim-up control (Sc); Swim up followed by Percoll control (SPc); Percoll and virus (Pv); swim-up and virus (Sv); and swim-up followed by a Percoll and virus (SPv). The Percoll gradient was prepared with 0.5 mL 90% under 0.5 mL 45% Percoll solution, centrifuged at 200 × g for 10 min; the pellet was collected and washed with Sperm-TALP. The swim-up was prepared with 0.3 mL of semen under 1 mL of Sperm-TALP. After 60 min of incubation at 38.5°C, 0.8 mL of the supernatant was collected and centrifuged at 200 × g for 10 min. The combination of techniques was prepared first with swim-up, and then the swim-up pellet was submitted to a Percoll, as previous described. The final pellet of each separation procedure was evaluated for sperm motility, sperm concentration, and virus isolation in cell culture. We used Student’s t-test for statistical analysis (P < 0.05). Preliminary results showed no significant statistical difference between groups of the same sperm separation techniques when analyzing sperm motility (Pc = 90%; Sc = 90%; SPc = 90%; Pv = 80%; Sv = 90%, SPv = 90%) and sperm concentration (million mL-1; Pc = 28; Sc = 11; SPc = 3; Pv = 18, Sv = 11, SPv = 4). The virus isolation in cell culture presented the following results (3 replicates): Pv = 103.92a, 103.8b, 102.46c TCID50 mL-1; Sv = 103.62, 103.13, 102 TCID50 mL-1; and SPv = undetectedd, undetectede, 102.92f TCID50mL-1 (different letters for means indicate statistical significance). The control groups of all techniques presented a cytotoxic effect probably because of the contact with sperm, Percoll, and sperm-TALP residues in the analyzed sample. Because of those cytotoxc effects in the cell culture, molecular biology techniques such as RT-PCR will be used as a complementary test to confirm the presence of BVDV in the samples. A large number of repetitions will be performed for a better statistical analysis. However, these preliminary results showed that the combination of swim-up followed by Percoll promoted a significant reduction in the number of viral particles in the semen samples compared with Percoll alone, considering that in 2 of the 3 repetitions it was not possible to detect the virus in cell culture. CNPq, CAPES.