Elevated fluctuating levels of bilirubin are a common problem in clinical studies. Differentiation between a drug-related adverse event and the diagnostic symptom for Gilbert's syndrome (GS), an idiopathic unconjugated hyperbilirubinemia, is more or less impracticable since the diagnosis of GS is by exclusion. The aim of this investigation was to evaluate the correlation of unspecific elevated bilirubin levels and the occurrence of GS with a described polymorphism in the uridine diphosphat glucuronosyltransferase 1A1 (UGT1A1) in a predominately Caucasian population. 304 volunteers (152 male, 152 female) were genotyped for the UGT1A1 promoter polymorphism by PCR amplification and polyacrylamide gel electrophoresis. Serum bilirubin levels and liver enzymes were determined and GS was diagnosed using clinico-chemical criteria. 23/13 subjects displayed the homocygote variant, 73/66 the heterozygote variant and 56/72 wildtype (male/female, respectively). 23 male and 3 female volunteers fulfilled the clinical criteria for GS (15.1, respectively 2.0%). Men exhibited higher serum bilirubin levels than women with a mean (SD) of 14.37 (8.92) micromol/l compared to 10.17 (5.37) micromol/l, respectively (p < 0.001). The homocygote mutant promoter length correlated well with serum bilirubin levels and with the clinical diagnosis of GS (p < 0.001 each). Genotyping of the UGT1A1 promoter polymorphism is a cheap and unequivocal method for predicting elevated and fluctuating bilirubin levels. It is better suited to this purpose than the clinical diagnosis which is based on exclusion. The genotyping of UGT1A1 promoter polymorphism can help to improve safety and the reliable assessment of adverse events in clinical studies. Our data additionally support the demand to refine the bilirubin reference values.