Abstract Over a period of four years, sodium and potassium chloride at 1 and 2 cwt/ac/annum, and mixtures in various proportions, were compared on a grass-clover pasture in a field trial on Horotiu sandy loam, using the technique of mowing with return of clippings. Responses to NaC1 were much smaller than those from the same rates of KC1. Yield increases of mixed herbage from 1 and 2 cwt NaC1 averaged 13 per cent and 18 per cent respectively in the first two years, but fell below 10 per cent and were no longer significant in the third and fourth years. In the presence of sub-optimal amounts of KC1, however, 1 cwt NaC1 continued to give responses throughout the trial period. Responses to 1 and 2 cwt KC1 averaged 32 per cent and 49 per cent respectively over the four years. Two cwt NaC1 increased the percentage of white clover (Trifolium repens L.) by 25 per cent and the per acre yield by 41 per cent, but 2 cwt KC1 increased these values by 166 per cent and 295 per cent respectively. Ryegrass (mostly Lolium perenne L.) percentages were little affected, but 2 cwt NaC1 increased ryegrass yields by 25 per cent, and 2 cwt KC1 by 78 per cent. All treatments reduced the percentage of other grasses, but had little effect on their yields. Sodium chloride increased plant K levels in all samples taken in 1954 and 1955 when K deficiency was slight to moderate. In 1957-8, however, when K deficiency in the plots without K was acute, such increases were found only in the moderately deficient plots which received sub-optimal amounts of KC1. Soil analyses failed to demonstrate release of non-exchangeable K from the soil by NaC1 in the absence of plants. The NaC1 responses appear to be best explained as due mainly to increased K uptake or displacement of K from the roots to the foliage.