Many epidemiological studies assess nutritional status based on single blood measurements, without verifying if these remain reliable over repeated measurements. This study assessed the reliability over a period of 2 to 5 yr of plasma carotenoids, vitamin C, retinol, tocopherols, and serum compounds involved in 1-carbon metabolism in a subsample of Dutch participants of European Prospective Investigation into Cancer and Nutrition (EPIC). Blood samples from 38 men from MORGEN-EPIC and 35 women from Prospect-EPIC were collected between 1993 and 1997 and again after 2 to 5 yr. The reliability of plasma carotenoids, retinol, vitamin C, and tocopherols, and of serum folate, homocysteine, and vitamins B6 and B12 was estimated using an intraclass correlation coefficient (ICC). Serum homocysteine and vitamin B12 were highly reliable biomarkers, with ICCs of 0.91 and 0.75, respectively. All other analyzed biomarkers had a slight or fair reliability over several years (ICCs ranged from 0.17 to 0.56). Most examined biomarkers showed reliability values that may lead to considerable attenuation of the risk estimate when used as exposure assessment in a risk model. If multiple measurements are not available, the risk estimates can be adjusted for the regression dilution using the ICC as adjustment coefficient.