In 14 retired lead workers, followed for over 18 years after end of exposure, repeated analyses of lead levels in finger bone by an in vivo X-ray fluorescence method revealed a decrease of lead concentration. The data were analysed using an exponential retention model. For the whole group the biological half-time was 16 (asymptotic 95% confidence interval, CI 12,23) years. The median of the estimated bone lead levels at the end of exposure was 85 micrograms.g-1 above the "background" (3 micrograms.g-1). A simultaneous follow-up of blood lead levels displayed a decrease, which could be described by a tri-exponential retention model with group half-times of 34 (CI 29,41) days, 1.2 (CI 0.9,1.8) years, and 13 (CI 10,18) years, respectively. The median of the estimated blood lead levels at the end of exposure for the three components were 0.49, 0.61, and 1.1 mumol.l-1 above the "background" (0.38-0.56 mumol.l-1), respectively. The well-documented decrease of lead exposure in the general population over the years, urged the use of a decreasing "background" of blood lead during the time of the study. The slowest of the three components represented the skeleton (probably mainly cortical bone), as did mainly probably also the intermediate one (trabecular bone). The data show the rather slow turnover of lead in the skeleton, the usefulnes of in vivo skeletal lead measurements as a long-term exposure index, and the importance of bone as a source of "endogenous" lead exposure.