About one fourth of patients with bipolar disorders (BD) have depressive episodes with a seasonal pattern (SP) coupled to a more severe disease. However, the underlying genetic influence on a SP in BD remains to be identified. We studied 269 BD Caucasian patients, with and without SP, recruited from university-affiliated psychiatric departments in France and performed a genetic single-marker analysis followed by a gene-based analysis on 349 single nucleotide polymorphisms (SNPs) spanning 21 circadian genes and 3 melatonin pathway genes. A SP in BD was nominally associated with 14 SNPs identified in 6 circadian genes: NPAS2, CRY2, ARNTL, ARNTL2, RORA and RORB. After correcting for multiple testing, using a false discovery rate approach, the associations remained significant for 5 SNPs in NPAS2 (chromosome 2:100793045-100989719): rs6738097 (pc = 0.006), rs12622050 (pc = 0.006), rs2305159 (pc = 0.01), rs1542179 (pc = 0.01), and rs1562313 (pc = 0.02). The gene-based analysis of the 349 SNPs showed that rs6738097 (NPAS2) and rs1554338 (CRY2) were significantly associated with the SP phenotype (respective Empirical p-values of 0.0003 and 0.005). The associations remained significant for rs6738097 (NPAS2) after Bonferroni correction. The epistasis analysis between rs6738097 (NPAS2) and rs1554338 (CRY2) suggested an additive effect. Genetic variations in NPAS2 might be a biomarker for a seasonal pattern in BD.