Determining valid zygosity is a basic and important requirement in a twin study, because misdiagnosing zygosity leads to biased results. The Healthy Twin Study has collected data from adult like-sex twins and their families since 2005. In the study, a questionnaire to determine zygosity was developed comprising four questions; one concerning the degree of resemblance, and three concerning the degree of confusion by the resemblance. Among 2,761 individuals (624 twin pairs) of twin and their families, 406 pairs of twins (mean age 38.3, 63.5% women) with both questionnaire and genotype information were selected to examine the validity of the zygosity questionnaire using 16 short tandem repeat markers. We first determined individual zygosity including undetermined category, and then decided the zygosity of a twin pair using a decision tree. Sensitivity of questionnaire diagnosis was 98.8% for monozygotic (MZ) and 88.9% for dizygotic (DZ) twins, and positive predictive value was 97.2% for MZ and 95.0% for DZ. When we compared correctly and wrongly diagnosed twin pairs, misdiagnosed DZ twins (nine pairs) showed striking similarity in stature or obesity even exceeding that of true MZ twins. Our finding suggests that a parsimonious questionnaire method of diagnosing the zygosity will be useful, and adding physical or physiological measurements to a questionnaire of zygosity diagnosis will either confound the correct diagnosis or reduce the efficiency of the study compared with using questionnaire alone or with introducing genotyping.