Two novel approaches to the analysis of twin data are illustrated with data from birth weight in twins. First, two possible covariates of birth weight are fitted to the data simultaneously, allowing for linear effects of these variables, and their correlation. Second, information on chorionicity is used to estimate the effects of chorion type on birth weight. The data were collected from a large sample of twins born in East Flanders, Belgium. Variation and covariation in twins were considered as a function of sex, chorionicity, maternal age, gestational age, and genotype. No evidence for sex differences in causes of variation was found. As expected, the largest source of variation in bith weight was associated with gestational age. Other common environmental influences were non-significant. Heritability was significant, constituting approximately 40% of variation not associated with maternal and gestational age. A small but significant effect of chorionicity was found, such that dichorionic twins show a greater similarity than monochorionic.