Although the retinoblastoma gene has been isolated and sequenced, the difference in penetrance and expressivity among families has not yet been fully explained. Balanced chromosomal insertion involving the 13q14 regions has been shown to account for some families with several unaffected carriers. Since there could be cases with karyotypically undetectable insertions, we tested whether this mechanism was general enough to explain the whole difference in expressivity among families. Using 166 pedigrees, reported in nine series available in the literature (including our own), we conclude that balanced insertion cannot entirely explain the familial data, even if we allow for a reduced viability of unbalanced gametes. Other mechanisms are proposed and discussed in this paper.