A greenhouse experiment was carried out from January–May 2001, to evaluate the effects of elemental sulphur inoculated with Thiobacillus, compared with gypsum, in the amendment of two saline sodic soils (Neosol Fluvic Salic sodic) from the Brazilian semi-arid region, and on growth of the tropical legumes cowpea and yam bean, inoculated with specific rhizobia strains. The treatments consisted of sulphur rates (0·6, 1·2 and 1·8 t/ha) and gypsum (1·8 and 3·6 t/ha), and irrigation water containing the salts NaHCO3, MgCl2, CaCl2, NaCl and KCl, with electrical conductivity 0·2 dS/m at 25 °C. There was a treatment with no sulphur or gypsum applied. The correctives increased Na+ and electrical conductivity on leached solution, and decreased soil pH and exchangeable cations, especially Na+. Sulphur inoculated with Thiobacillus was more efficient than gypsum, reducing soil-exchangeable sodium. Sulphur with Thiobacillus in Soil 1 reduced pH (8·2 to 4·7) and electrical conductivity of the soil saturation extract (15·3 to 1·7 mS/cm) to values below those used for classification as saline and sodic soil. The growth of the tropical legumes cowpea and yam bean was increased by rhizobia inoculation when soil ameliorants were used, especially sulphur in combination with Thiobacillus.