Anticipation was investigated in schizophrenia (SZ) and bipolar disorder (BP) while addressing several biases in 18 large families (154 subjects) from Eastern Que ´bec densely affected by SZ, BP, or both over three generations. In particular, we controlled for an information bias using a measure of quality and quantity of clinical information (QOI) concerning the subjects' illness. Otherwise, spurious anticipation could have arisen because we found that QOI varied with the generations as well as with the severity of illness. Although anticipation was investigated separately for SZ and BP, both disorders were also included in one analysis that tested anticipation under the unitary hypothesis that the SZ and the BP spectrums represent a continuum of severity of the same disease. Age of onset (AOO) and five indices of severity were tested for anticipation. Two statistics were used: the difference in the mean AOO or severity between two successive generations, and the mean difference in parent-offspring pairs (POP). The study led to four main findings: 1) the choice of the statistics greatly influenced the results, POP yielding systematically greater biased estimates; 2) for SZ and BP, the evidence for anticipation with the five severity indices vanished after controlling for QOI; 3) as regards AOO a decrease of 8.6 years, p = 0.0001, and 5.3 years, p = 0.009 in AOO was found for SZ between Generations 1-2, and 2-3, respectively, despite controlling for QOI and addressing all biases; and 4) conversely for BP, anticipation with AOO may be due to censoring.