Observational studies addressing the relation between selenium and human health, particularly cancer risk, yielded inconsistent results, while most recent randomized trials showed a fairly consistent pattern suggesting null or adverse effects of the metalloid. One of the most plausible explanations for such inconsistencies is inadequate exposure assessment in observational studies, commonly carried out by measuring total Se content without taking into account the specific exposure to the individual chemical forms of the metalloid, whose toxic and nutritional properties may vary greatly. Data on the distribution of these species in human blood and their correlation with overall selenium levels are very limited. The concentrations of organic and inorganic selenium species were analyzed in serum of fifty subjects sampled from the general population of the municipality of Modena, northern Italy, aged from 35 to 70 years. Samples were collected during a 30-month period, and determinations of selenium species were carried out using high pressure liquid chromatography coupled with inductively coupled plasma dynamic reaction cell mass spectrometry. The majority of selenium was found to be present as organic species, but the inorganic forms showed higher levels than expected. These species showed limited correlations with age, sex and body mass index, while the organic forms increased in subjects consuming selenium-containing dietary supplements and decreased in smokers. The length of the sample storage period strongly influenced the distribution of selenium compounds, with a clear tendency towards higher inorganic and lower organic selenium levels over time. In multivariate analysis adjusting for potential confounders, total serum selenium correlated with human serum albumin-bound selenium and, in males, with two organic species of the metalloid (selenocysteine and glutathione peroxidase-bound selenium), while little association existed with the other organic forms and the inorganic ones. These findings highlight the potential for exposure misclassification of observational epidemiologic investigations based on overall selenium content in blood and possibly other tissues, and the critical role of the storage conditions for speciation analysis.