Several reports have suggested the presence of anticipation in bipolar affective disorder (BPAD). In addition, independent studies using the RED (repeat expansion detection) have shown association between BPAD and longer CAG/CTG repeats. Therefore loci with large CAG/CTG repeats are plausible candidates in the inheritance of BPAD. The present study assesses the length of the repeats in four loci: the ERDA-1 locus which is known to account for most of the long CAG repeats detected by RED, the SEF2-1b locus which is placed in a region where positive linkage results have been reported and the loci MAB21L and KCNN3 as functional candidate genes. A Brazilian case-control sample with 115 unrelated BPAD patients and 196 healthy control subjects and 14 multiply affected bipolar families was investigated. With the case-control design the distribution of alleles between the two groups did not approach statistical significance. The extended transmission disequilibrium test (ETDT) performed in our families did not show evidence for linkage disequilibrium. Parametric and non-parametric linkage analysis also did not provide support for linkage between any of the four loci and BPAD. Our data do not support the hypothesis that variation at the polymorphic CAG/CTG repeat loci ERDA-1, SEF2-1b, MAB21L or KCNN3 influence susceptibility to BPAD in our sample.