A mailed questionnaire dealing with similarity in childhood, and how often and by whom the twins were mixed up, was applied to 290 same-sex adult twin pairs. For about 75% of pairs ten genetic markers were analysed. The agreement in zygosity classification between questionnaire and blood typing was high. Irrespective of whether raw scores or discriminant function analysis was applied, the agreement was nearly 95%. When a decision tree was applied, the percentage rose to nearly 96%. Considering that probably four of the twin pairs were wrongly classified as MZ by genetic markers, the percentage rose to 96 and 98, respectively.