Soil samples contaminated in the laboratory with a Nacl solution were leached with water after various amendments had been added. with no amendments added, percolation rates were reduced more on a light-textured Dark Brown chernozemic A than on a medium-textured Black chernozemic A. Undisturbed B horizon cores showed a smaller decrease in percolation rate than the A horizons, but natural gypsum or carbonates were of no benefit in maintaining percolation through the unamended contaminated subsoils. Percolation g"n".uily increased as more ca-amendment was added to the contaminated A horizons, but rate of (surface-applied) amendment had no effect on the percolation throush the B horizon cores. Gypsum mixed into the contaminated soll was much more-eflective than gypsum applied on the surface; adding NH1NO, to the gypsum had no effect. The efficiency of incorporated amendments decreased in the order: gypsum > MgSO, > Ca(NO,)r. Surface-applied Ca(NOr). was a better amendment than mixed-in Ca(NO.,)r. kaching losses decreased in the order: Cl > Na > Mg > Ca > K. Initial losses of Cl, Na, Mg and Ca were faster than expected from the rule of thumb that one pore volume of water reduces the salt content by about one-half. Subsequently, losses were much slower than in the rule of thumb as most of the remaining cations are in exchangeable form.