A retrospective exposure assessment of 1269 study subjects was completed for use in a multi-site case-control study of the relationship between protracted workplace external radiation exposure and leukaemia mortality. The majority of exposure data result from film badge monitoring programmes at the four US weapons production facilities and a US Naval shipyard. Bias and uncertainty in reported exposures among study facilities and across time were as result of differences in incident photon energy, exposure geometry, dosemeter type and dosimetry methods. These sources of measurement uncertainty were examined by facility and time to derive bias factors (B) for normalising exposures. In conjunction with facility reported results, the bias factors provide a means to estimate the equivalent dose, penetrating to a depth of 10 mm [H(p)(10)] and the equivalent dose to the active bone marrow for use in the epidemiological study. Uncertainty was expressed as the constructed 95% confidence interval (i.e. the 2.5th-97.5th% range) of the estimated parameter. The bias factors indicate that recorded exposures provide a reasonable estimate of H(p)(10) (bias factor near unity) and overestimate equivalent dose to active bone marrow (H(T)) by a factor between 1.2 and 1.7. On average, dosemeter-response uncertainties estimated using Monte Carlo simulation were approximately +/-19 and +/-33% for H(p)(10) and H(T), respectively.