A layer of Ca(H2PO4)2.H2O (MCP) or MCP plus CaSO4.2H2O was spread over duplicate columns of six soils to simulate the effects of banded MCP or superphosphate (MCP plus CaSO4.2H2O) on soil solution composition. A separate column was set up without fertilizer addition for each soil to act as a control (background) treatment. The soils used were 0-10 cm samples from two Kurosols, a Ferrosol, a Vertosol, a Kandosol, and a 50-60 cm sample from the Kandosol. Prior to fertilizer addition, the columns were wet up to the water content at a matric suction of 10 kPa. Following 5 days of fertilizer-soil contact, soil sections were recovered at 5 mm increments from the fertilizer layer to a distance of 50 mm. Soybean (Glycine max: (L.) Merr.) seedlings were grown for 48 h in each section and relative root elongation (RRE) was determined. Soil solution was then extracted from each section and analysed. The distance of phosphorus (P) movement from both MCP and MCP plus CaSO4.2H2O was better correlated with P buffer capacity determined at a solution P concentration of 3.2 �M than at 320 �M. This suggests that the precipitation reactions which occur at the fertilizer site when MCP dissolves are independent, of the soil, and it is only in soil sections further removed from the fertilizer source (i.e. with lower soil solution P concentrations) that the P sorption properties of the soil become important in determining the extent of P movement. The amount of inorganic P (Pi) in the soil solution was summed over all soil sections for each fertilizer source, and was correlated with citrate-dithionite extractable Fe and Al using step-up regression techniques. Citrate-dithionite extractable Fe was highly correlated with P-i (r = -0.937, P < 0.001), and the addition of citrate-dithionite extractable Al did not significantly (P = 0.05) increase the variation accounted for. RRE decreased in proximity to the fertilizer. When RRE was plotted against the electrical conductivity of the soil solution, all data points fell below the regression line previously obtained for various salts (Moody et al. Aust. J. Soil Res. 1995, 33, 673-87), indicating that the reduction in RRE was not due solely to osmotic effects. Multiple regression analysis indicated that a combination of the activities of Al3+ (aAl) and Mn2+ (aMn) explained 83% of the variation in RRE when both fertilizer sources were considered in all soils except the Kurosols. There was evidence of organic complexing of soil solution Al in the two Kurosols and so an accurate estimate of Al3+ activity could not be made. For the soils other than the Kurosols, separate regressions of RRE against ant and a(Mn) indicated a 10% reduction in RRE set activities of 1.9 and 70 �M, respectively. Based on these activities, banding of MCP and MCP plus CaSO4.2H2O caused Al toxicity in all soils, and Mn toxicity in all soils except one of the Kurosols. Manganese toxicity occurred further from the fertilizer band than Al toxicity in the Ferrosol and the Kandosol. The dual occurrence of Al and Mn toxicities indicates that both factors need to be considered simultaneously when determining the effects of banded fertilizer on RRE.