The analysis of surveillance capsule dosimetry has played a defining role in the database used for trend curve development. The extent to which calculational meth ods and nuclear data have improved over the past 30 years suggests that early analyses may be significantly biased. Accordingly, the comparability of fast fluence estimates (f) made nowadays to those of the late 1960s and early 1970s may be affected, with possible consequences for the trend curves fit to these data.