Since fuel oil ash contains vanadium (V), the measurement of urinary levels of V may provide a biological marker in workers exposed to fuel oil ash. The usefulness of urine V samples as a biological monitoring tool ultimately depends on determining the appropriate time of sampling relative to when exposure occurs. Twenty boilermakers were studied during the overhaul of a large oil‐fired boiler. A total of 117 urine samples were collected, 65 start‐of‐shift (S‐O‐S) and 52 end‐of‐shift (E‐O‐S) samples. Air V exposures were estimated with personal sampling devices and work history diaries. Air V concentrations ranged from 0.36 to 32.19 μg V/m3, with a mean ±SD of 19.1 ± 10.7, and a median of 18.5. On the first day of work on the overhaul, the V urine levels at the E‐O‐S (mean ±SD were 1.53 ± 0.53, median was 1.52 mg V/g creatinine) were significantly higher than those at the S‐O‐S (0.87 ± 0.32, median was 0.83), P = 0.004. However, the V concentrations of the S‐O‐S urine samples on the last Monday of the study were not significantly different from the S‐O‐S urine levels on the previous Saturday, a time interval of about 38 hr between the end of exposure and sample collection. The Spearman correlation coefficient (r) between the S‐O‐S urine V and the workplace concentration of V dust during the previous day was r=0.35. In summary, the results suggest a rapid initial clearance of V (elevating the E‐O‐S V concentration on the first day of work relative to the S‐O‐S concentration), followed by a slow clearance that is not complete 38 hr after the end of exposure, as evidenced by the Monday morning urine V concentrations. The Spearman correlations suggest that the S‐O‐S urine is preferred to the E‐O‐S urine for across‐shift biological monitoring of V exposure.Am. J. Ind. Med. 33:55–60, 1998. © 1998 Wiley‐Liss, Inc.