This paper describes a blind cross-over study on the gastrointestinal uptake of soluble and insoluble cobalt compounds (8.5 mumol/day) in 12 male and 11 female volunteers. In a controlled study it was found that the gastrointestinal uptake of the soluble cobalt compound cobalt chloride was considerably higher than the uptake of the insoluble cobalt compound cobalt oxide (urine ranges: < 0.17-4373 and < 0.17-14.6 nmol/mmol creatinine, respectively). Surprisingly, it was shown that ingestion of controlled amounts of soluble cobalt compound resulted in significantly higher urinary cobalt levels (P < 0.01) in females (median: 109.7 nmol/mmol creatinine) than in males (median: 38.4 nmol/mmol creatinine). The results suggest that the gastrointestinal uptake of cobalt is higher for females than males. The present study shows that the normal levels of cobalt in blood and urine in a non-random-selected group of Danes are low. As the fraction of values below the detection limit of the analytical method was 0.19 and 0.33 for urinary cobalt in females and males, respectively, distribution-free one-sided tolerance intervals were chosen to describe the values. The precision of the estimate of the tolerance intervals was expressed as coverage intervals. In females the 95% one-sided tolerance limit calculated for cobalt in blood and urine was 8.48 and 55.10 nmol/l with coverage intervals of 90% +/- 6.5% and 95% +/- 4.2% at a probability of 0.95, respectively. Even though the studied groups of males and females were not representative for the general population, the study indicates that oral exposure may be important in occupational settings.(ABSTRACT TRUNCATED AT 250 WORDS)