Nine field experiments with sugar beet in 1968–70 tested eight amounts of nitrogen fertilizer (0–290 kg N/ha) on a shallow calcareous loam (Icknield Series), on a deep sandy loam (Newport Series) and on a heavy clay loam (Evesham Series). The amount of mineral nitrogen in the top and sub-soils was determined before applying fertilizer and at monthly intervals from May to October in plots given 0, 125 and 250 kg N/ha. The crop on these plots was also sampled at monthly intervals throughout the growing season and the yield and nitrogen uptake determined. The soil analyses indicated that in springs with average rainfall, the leaching losses of nitrogen fertilizer are negligibly small, although there was some evidence that losses may be greatest on sandy loams. In very wet springs such as 1969, with almost double the normal rainfall, losses through leaching are considerable – on average, 40 kg N/ha. Dry-matter yields and response to nitrogen fertilizer differed between the three soils consistently from year to year. On the calcareous loam, neither amount of fertilizer changed the dry-matter yield of roots in any year. The crop on the clay loam needed a small dressing and on the sandy loam a larger dressing of fertilizer for maximum root dry-matter yield. Uptake of nitrogen by the crops usually paralleled the decreases in soil mineral nitrogen although on the clay loam nearly a third of the nitrogen applied could not be accounted for in the soil or plants, suggesting that some denitrification may have taken place. When the amount of nitrogen taken up by unfertilized crops is allowed for, the percentage recovery of applied fertilizer nitrogen at final harvest ranged from 42% on the calcareous loams to 62% on the sandy loams.