Three lime treatments (0, 5, and 10 tonnes CaCOa/ha) were applied to "undisturbed" soil cores from the three soil types (topsoil pH approx. 5.5). Half the cores from each lime treatment were sown to lucerne and half left fallow. Lucerne was harvested over a 2-year period and analysed for Ca, S, P, and K; some harvests were also analysed for B, Zn, Cu, Mo, Mn, and Fe. After 2 years all cores were dismantled and depth increments analysed for pH, exchangeable Ca, Mg, and K, P-retention, "water-soluble"P, and adsorbed sulphate. With the exception of "water-soluble" P, results were subjected to separate four-way analyses of variance for each soil, examining core, depth, plant, and lime main effects and their interactions. A smaller number of samples were analysed for "water-soluble" P and a different analysis of variance, combining the data from all three soils, had to be applied to these results. The data and analyses of variance for pH and exchangeable Ca showed that most lime applied had been retained in the top 15-20 em of the three soils, with little further penetration in Kiwitea silt loam and Marton silt loam, but some penetration to 60 cm in Dannevirke silt loam. This suggested different water movement properties in Dannevirke silt loam, leading to more leaching. There was a significant lime and depth interaction for "water-soluble" P. Although the reason for this was not obvious because of the variability of the results, it was suggested that the interaction reflected an increase in P solubility caused by liming of the order of 70% in the surface layers of the three soils. There was little apparent effect of lime on adsorbed sulphate, P-retention, exchangeable Mg, or exchangeable K. Analyses of variance applied to plant data showed a significant effect of lime increasing yield with Marton silt loam, but decreasing yield and %K content with Kiwitea silt loam at certain harvests. Lime had no significant effects on P or S contents of lucerne. Stepwise multiple regression analyses of yield against Ca, S, P. and K content gave no significant relationships for any of the three soils. The microelement determinations indicated that where lime had an effect, it was in the established direction for that element. Microelements did not appear to be deficient under any treatment. however, and therefore appeared unlikely to explain the observed yield differences.