The transmission disequilibrium test (TDT) is widely used as a robust statistical method to test for genetic association due to linkage based upon analysis of parent-proband trios. The TDT and other family-based tests (eg haplotype relative risk method) are commonly used in association studies including those of ADHD because of concerns that the case-control design has a strong tendency for false positives due to poor matching between cases and controls. Unfortunately, it is not always possible to obtain DNA from both parents in studies of this design, even where the onset of disorder is in childhood, and usually the missing parent is the father. Despite the fact that methods exist for analysis where one parent is missing, many family-based studies are based on the collection or analysis of complete trios only. However this selection process might potentially introduce bias, particularly for studies of behavioural phenotypes like ADHD because the phenotype of proband or parents might influence family stability and therefore complete parental ascertainment. We set out to examine whether children with ADHD and for whom DNA samples from fathers were missing ('duos') differed phenotypically from children for whom genotype information was available from both parents ('trios'). Children from duos showed a significantly higher frequency of DMS-IV ADHD-combined type, significantly more co-morbid conduct disorder and conduct disorder symptoms, and a trend for higher total ADHD symptom scores. Excluding duos from sample collection and analysis may result in systematic bias. If comorbid conduct disorder and ADHD-combined type index increased genetic liability, exclusion of duos could further reduce the power of the TDT (and similar tests) to detect susceptibility genes for ADHD, or replicate effects detected by case-control analysis.