The purpose of this study was to examine potential biases in family history reports of problem gambling and gambling frequency. Same-sex twin pairs discordant for a history of problem (n = 230 pairs) and pathological gambling (n = 48 pairs) and for three indexes of gambling frequency (ever gambling, monthly gambling, and weekly gambling; n = 44–517 pairs) were identified from a large Australian national twin study. The problem gambling affected twin was significantly more likely to endorse paternal problem gambling than the problem gambling unaffected cotwin (OR = 5.5), and similar findings were obtained for family history reports of gambling frequency (OR = 2.0–2.8). These results could not be explained by differences between the discordant pairs in whether they had spent time gambling with the parents; there was no association between a history of problem, monthly or weekly gambling and having gambled with the parents among discordant twin pairs. The results of this study suggest that relying solely on family history assessments of disordered gambling and gambling involvement can lead to incorrect estimates of the strength of the family history effect.