The 25-hydroxy vitamin D (25(OH)D) production caused by UVB exposure is usually underestimated as the concurrent degradation of 25(OH)D is not considered. Therefore, the decrease in 25(OH)D was investigated during a 7-week period in winter when ambient UVB is negligible. Twenty-two healthy Danish individuals (113 samples) participated and had a mean and steady maximal 25(OH)D start level of 132 nmol l(-1) (range of 68-216 nmol l(-1)) due to long-term UVB treatment prior to this study. In this group with high 25(OH)D start levels, the decrease in 25(OH)D was best described by an exponential model. This suggests a quantitatively larger elimination of 25(OH)D at high 25(OH)D start levels. A linear model (logarithm of 25(OH)D) including personal start levels as intercepts and a slope influenced by gender and the vitamin D receptor gene polymorphism rs2228570 explained 87.8% of the observed variation. The mean half-life was 89 days with a difference in half-life of 120 days between a male with rs2228570 genotype GG (59 days) and a female with rs2228570 genotype AA/AG (179 days). Thus, these two parameters explained a large part of the observed inter-individual variation of 25(OH)D. Furthermore, the decrease was analysed in two groups with medium and low 25(OH)D start levels resulting in longer half-lives of 149 days and 199 days, respectively. The longer half-lives at lower 25(OH)D levels may be caused by storage mobilisation, changed catabolism or increased intestinal absorption.