Sleep duration is genetically regulated, but the genetic variants are largely unknown. We aimed to identify such genes using a genome-wide association study (GWAS) combined with RNA expression at the population level, and with experimental verification. A GWAS was performed in a Finnish sample (n=1941), and variants with suggestive association (P<5x10(-5)) were tested in a follow-up sample from the same population with sleep duration (n=6834) and time in bed (n=1720). Variants with pointwise association of P<0.05 in the follow-up sample were analysed further. First, we correlated genotypes with transcript expression levels with sleep duration (n=207). The expression levels of significant transcripts were further studied in experimental sleep restriction. Of the 31 variants with P<5x10(-5) in the discovery sample, three variants showed nominal allelic association (P<0.05) in the follow-up sample: rs10914351, near PTPRU (P= 0.049), rs1037079 in PCDH7-CENTD1 (P=0.011) and rs2031573 near KLF6 (P=0.044). The risk alleles for shorter sleep (rs2031573 and rs1037079) were also associated with higher KLF6 and PCDH7 expression levels (P<0.05). Experimental sleep restriction increased the expression of KLF6 (P<0.01). These data suggest that rs2031573 near KLF6 or related loci and rs1037079 between PCDH7-CENTD1 or related loci may contribute to the regulation of sleep duration via gene expression. These results illustrate the utility of combining different analytical approaches to identify genetic determinants for traits related to sleep physiology. However, additional studies are needed in order to understand the roles of KLF6 and PCDH7 in sleep regulation.