Merino ewes have been selected, over 18 generations, for calm (C) or nervous (N) temperament using using an arena test and an isolation box test. We investigated the relative contributions of genotype versus the post-partum behaviour of the dam on the temperament of the lambs using a cross-fostering procedure. Forty-eight multiparous calm and 52 nervous ewes were artificially inseminated with the semen of a sire of the some temperament. At birth, 32 lambs of a given temperament line were cross fostered to ewes from the other line (16 N x C, 16 C x N), 34 lambs were cross fostered to ewes from the same line (15 C x C, 19 N x N) and 30 lambs were left with their birth mother (15 C, 15 N), to control for the effect of cross fostering. The temperament of the progeny was assessed at two occasions, one week after birth by measuring locomotor activity during an open-field test and at weaning (16 weeks) by measuring locomotor activity during an arena test and agitation score measured during an isolation box test. There was a genotype effect but no maternal or fostering effect on the tomb temperament at one week. This may be because the maternal behaviour of the foster ewes did not differ considerably between the calm and nervous mothers during adoption or within the first week, post partum. Similarly, at weaning, only a genotype effect was found on the locomotor and agitation score. Therefore, it appears that temperament in Merino sheep is mainly determined by the genetic transmission of the trait across generations rather than behaviours learned from the mother.