Thirteen field experiments were made during 6 years on seed-potato growing farms in south-east Scotland, comparing three rates, 70, 140 and 280 kg K/ha as potassium chloride and four rates of 0-54 kg Mg/ha as kieserite in 3 x 4 factorial experiments. More than 70 kg/ha of potassium increased total yield only at one site, which had a recent history of low K application and very low available K. Yields were decreased by more than 70 kg/ha of potassium at three sites in a season with abnormally dry conditions just after planting. Applied magnesium had little effect on total yield. Extra potassium increased ware yield but decreased seed yield, both consistently, probably because the fertilizer damaged some stolons at or before tuber initiation and fewer tubers developed. The ware/seed ratio was generally increased by extra K, but after the dry spring this ratio was unaltered or reduced. Applied magnesium had little effect on the ratio. Our results suggest that the present rates of K fertilizer, used for commercial seed-potato crops (114-138 kg K/ha) in south-east Scotland are excessive, and a much lower rate of approximately 70 kg K/ha would be adequate except on very low K sites.