We examined whether the adherence to major dietary patterns at baseline of 5824 nondiabetic Danes (30-60 y) enrolled in the nonpharmacological Inter99 intervention predicted changes in fasting plasma glucose (FPG) and postchallenge 2-h plasma glucose (2h-PG) concentrations during a 5 y period and whether a potential association was dependent on baseline glucose tolerance status. Through principal component analysis, a score for a traditional dietary pattern (characterized by higher intakes of high-fat sandwich spreads, red meat, potatoes, butter and lard, low-fat fish, sandwich meat, and sauces) and a score for a modern dietary pattern (characterized by higher intakes of vegetables, fruit, vegetable oil/vinegar dressing, poultry, pasta, rice, and cereals) were estimated for each person at baseline. Random effect models adjusting for relevant confounders were used to estimate changes in repetitive measures of FPG and 2h-PG. A higher modern score (of 1 SD) predicted an annual decrease in 2h-PG of 0.015 mmol/L (P < 0.01) regardless of glucose tolerance status. For individuals with isolated impaired glucose tolerance, a higher traditional score (of 1 SD) predicted an annual increase in 2h-PG of 0.083 mmol/L (P < 0.0001). In conclusion, glucose tolerance status did not, in general, affect the predictive effect of the dietary patterns. The study suggests that the risk of worsening 2h-PG concentrations may be smaller for individuals with a high modern dietary pattern score characterized by high intakes of vegetables, fruit, vegetable oil/vinegar dressing, poultry, pasta, rice, and cereals.