Genomewide scans of bipolar disorder (BP) have not produced consistent linkage findings. Follow-up studies using enlarged samples and enhanced marker density can bolster or refute claims of linkage and pave the way to gene discovery. We conducted linkage and association analyses, using a ~3-cM density map of 10 candidate regions, in a large BP pedigree sample (865 individuals from 56 pedigrees). The candidate regions were identified in a previous 10-cM genome-wide scan using a subset of this sample (373 individuals from 40 pedigrees). The present sample consists of the expanded original pedigrees ('core' pedigrees) and 16 additional pedigrees. We obtained experiment-wide significant linkage on chromosome 7q34 (LOD score 3.53, p<0.001), substantially stronger than that observed in the genome-wide scan. Support for linkage was sustained on chromosomes 2p13, 4q31, 8q13, 13q32, 14q21 and 17q11, though at a more modest level. Family-based association analysis was consistent with the linkage results at all regions with linkage evidence, except 4q an 8q, but the results fell short of statistical significance. Three of the previously implicated regions -9q31, 10q21 and 10q24 -showed substantial reduction in evidence of linkage. Our results strongly support 7q34 as a region harboring susceptibility locus for BP. Somewhat lesser, yet notable support was obtained for 2p13, 4q31, 8q13, 13q32, 14q21 and 17q11. These regions could be considered prime candidates for future gene finding efforts.