Recent genome wide association studies (GWAS) found associations between clozapine serum levels and single nucleotide polymorphisms (SNP) in intragenic region between cytochrome P450 1A1 (CYP1A1) and CYP1A2 and nuclear factor 1B (NFIB). The aim of this study was to perform another GWAS of polymorphisms associated with the serum levels of clozapine and norclozapine, their ratios, and to perform meta-analyses with two previous GWAS. Finnish clozapine patients (n = 170) with known smoking habits were genotyped. GWAS was performed with clozapine concentration/dose ratio (C/D), norclozapine C/D and clozapine/norclozapine-ratio as phenotypes, adjusting for age, sex and first four genetic principal components, and additionally for smoking and valproate use. The two other patient populations were from the British CLOZUK2 study (n = 2989) and Norwegian Diakonhjemmet Hospital study (n = 484). In the three population (n = 3643) meta-analyses the top SNP associated with the clozapine C/D ratio was rs2472297, located between the CYP1A1 and CYP1A2 genes. For the norclozapine C/D ratio, an association signal was found near uridine-5´-diphospho-glucunorosyltransferase (UGT) UGT2B10 gene. Additionally, rs3732218, an intron variant in the UGT1A family gene complex, was associated with norclozapine C/D. Lead SNP associated with the clozapine/norclozapine ratio was rs6827692, an intron variant near UGT2B7 gene. In the two population meta-analyses (n = 654) adjusting for smoking and valproate use, the UGT2B10 intron variant rs835309 was associated with the clozapine/norclozapine ratio. We suggest a UGT2B10 missense SNP rs61750900, in perfect linkage disequilibrium with UGT2B10 rs835309, as the probable causal variant. Our study confirms and extends the number of genetic variants affecting clozapine and norclozapine metabolism.