Reference intervals in clinical chemistry are commonly based on results of measurements in reference populations or are taken from the literature. A reference population should represent a defined group of individuals and be as similar as possible to the patients under investigation. Frequently, reference populations have been recruited from institutionalised healthy young people who do not necessarily fulfill these criteria. In the present study we describe the temporal changes in 37 commonly measured quantities in men and women from childhood to late in life. The samples were collected in the primary health care and sorted according to an assumed decision by the physician. The emerging group of individuals forms a reference population that was regarded as "non-diseased" and the results of measurements in this population are reference values. A remaining group of "non-healthy" were likewise identified for comparison. The central 95 percentile was wider than those usually assigned to the quantities whereas the medians almost coincided. In the "non-healthy" group the medians were shifted in a direction that would be expected from pathophysiology aspects.