To evaluate an adequate strategy for biological monitoring of aluminium (Al), a group of 62 Al welders (age in 1999: 23-51 years, median 35 years) was surveyed annually from 1999 to 2003 by determination of pre- and post-shift Al in urine and plasma. Biomonitoring was supplemented by personal air measurements of the total dust concentration. The welders' internal exposure was compared to the exposure of 60 non-exposed assembly workers (age in 1999: 21-51 years, median: 36 years) who were surveyed in 1999, 2001 and 2003. Having a nearly constant dust exposure, median concentrations of Al in urine (Al in plasma) of the welders decreased from 40.1 microg/g to 19.8 microg/g creatinine (8.7 to 4.6 microg/l). For the control group the median levels of Al in urine (plasma) ranged from 4.8 microg/g to 5.2 microg/g creatinine (2.4-4.3 microg/l) indicating a higher sensitivity for the marker Al in urine. No systematic differences have been found between pre- and post-shift internal exposure. This might be caused by the slow elimination kinetics and low systemic bioavailability of Al. A correlation analysis did not yield close relationships between dust exposure, Al in plasma and Al in urine underlining the importance of biomonitoring for assessment of Al exposure.