Heritability estimates based on two small cross-sectional studies in children indicate that the genetic contribution to individual differences in loneliness is approximately 50%. A recent study estimated the genetic contribution to variation in loneliness in adults to be 48%. The current study aims to replicate and expand these findings by conducting longitudinal analyses in order to study causes of individual differences in stability of loneliness throughout childhood. Univariate and multivariate longitudinal analyses are conducted in a large sample of young Dutch twins. Information on loneliness comes from maternal ratings on the Child Behavior Checklist. Using an average score of loneliness over ages 7, 10, and 12, results from the two previous studies are replicated and a heritability estimate of 45% is found. The remaining variance is accounted for by shared environmental influences (12%), and non-shared environmental influences (43%). The long-itudinal analyses, however, show that heritability is 58% at age 7, 56% at age 10, but drops to 26% at age 12. A parallel increase in influences of shared family environment is observed, explaining 6% of the variance at age 7, 8% at age 10 and 35% at age 12. The remaining variance is explained by relatively stable influences of nonshared environmental factors. Stability in loneliness is high, with phenotypic correlations in the range of 0.51-0.69. This phenotypic stability is mainly caused by genetic and nonshared environmental influences. The results indicate the importance of both innate as well as nonshared environmental factors for individual differences in loneliness. Further, different results between causes of individual differences for the average score of loneliness and results for age 12 from the longitudinal analyses, indicate the importance of longitudinal analyses with data at well-defined ages. © 2007 Wiley-Liss, Inc.