Field experiments were conducted at two locations in the seleniferous region of northwestern India from 2001 to 2006 to evaluate the efficiency of four cropping systems in removing Se from contaminated soil containing 2843–4345 μg Se per kg in the surface layer (0–15 cm). Rapeseed (Brassica napus) followed by arhar (Cajanus cajan), sunn hemp (Crotalaria juncea) or cotton (Gossypium arboretum) and wheat (Triticum aestivum) followed by rice (Oryza sativa) were the four cropping systems. The total biomass generated by Brassica-based systems ranged from 16 to 21 t/ha when harvested at maturity. Corresponding values for a wheat–rice sequence were 22–26 t/ha. Among the different crops at both the experimental sites, the highest Se content was recorded in leaves (157–209 mg/kg), grains (64–201 mg/kg) and stems (42–93 mg/kg) of Brassica and the lowest in the shoots (10–27 mg/kg), grains (5–13 mg/kg) and straw (13–20 mg/kg) of the rice crop. Except for S and P, concentrations of other nutrients (Zn, Cu, Mn and Fe) were not significantly affected by variations in the Se content of plants. Significant correlation coefficients were observed between Se and S (r = 0.838, P ≤ 0.001), Se and P (r = 0.817, P ≤ 0.001) at the peak flowering stage (n = 16), and r = 0.743, P ≤ 0.001 and r = 0.498, P ≤ 0.05, respectively, at the maturity stage (n = 16). Total Se removal through harvested biomass of rapeseed-based cropping sequences varied from 716 to 1374 g/ha/yr at peak flowering and 736–949 g/ha/yr at the maturity stage. Corresponding values for a wheat–rice system were 435–492 and 370–517 g/ha/yr, respectively. The amount of Se recycled through leaf senescence ranged from 255 to 500 g/ha/yr for Brassica-based cropping systems. In the wheat–rice system, Se addition through irrigation varied from 170 to 243 g/ha/yr and was three to four times more than that added in Brassica-based systems. On completion of the phytoremediation experiments at site I, Se removal through harvested biomass at maturity was 1.7–5.1% of total Se in the soil down to a depth of 120 cm and 4.8–13.2% at site II. Analysis showed that Se losses under different crop rotations were 18.5–24.5% at site I and 21–33% at site II of total soil Se. Thus, at both sites 16–20% of total Se lost from the soil was unexplained. Results show that Brassica-based cropping systems lead to significant reductions in Se capital of contaminated soil over 2–3 years. Although a long-term commitment is required, adoption of Brassica-based systems as a regular agricultural practice must lead to sustainable management of seleniferous soils.