Genetic variations in clock-related genes were hypothesized to be involved to in the susceptibility of mood disorders MD (both unipolar (UPD) and bipolar (BPD) disorders). In our work we investigated role of gene variants form four core period proteins: CLOCK, ARNTL, TIM and PER3. The total sample comprised from 744 mood disorders inpatients (UPD = 229, BPD = 515) and 635 healthy voluntary controls. The 42 SNPs from four genes of interest were genotyped. We used single polymorphisms, haplotypes, SNPs interactions and prediction analysis using classical statistical and machine learning methods. We observed association between two polymorphisms of CLOCK (rs1801260 and rs11932595) with BPDII and two polymorphisms of TIM (rs2291739, rs11171856) with UPD. We also detected ARNTL haplotype variant (rs1160996C/rs11022779G/rs1122780T) to be associated with increased risk of MD, BPD (both types). We established significant epistatic interaction between PER3 (rs2172563) and ARNTL (rs4146388 and rs7107287) in case of BPD. Additionally relation between PER3 (rs2172563) and CLOCK (rs1268271 and rs3805148) appeared in case of UPD. Classification and Regression Trees (C and RT) showed significant predictive value for 10 polymorphisms in all analyzed genes. However we failed to obtain model with sufficient predictive power. During analyses of sleep disturbances sample, we found carriers of homozygote variants (ARNTL: rs11022778 TT, rs1562438 TT, rs1982350 AA and PER3: rs836755 CC) showing more frequent falling asleep difficulties when compare to other genotypes carriers. Our study suggested a putative role of the CLOCK, TIM, ARNTL and PER3 and polymorphisms in MD susceptibility. In our analyses we showed association of specific gene variants with particular types of MD. We also confirmed necessity of performing separate analyzes for BPD and UPD patients. Comprehensive statistical approach is required even with individual symptoms analyses.