Radiation measurement results are typically stated as an estimated act
ivity concentration with an associated 95% error bound. Very frequentl
y, however, the error bound cited at the ''95% confidence level'' is n
othing more than counting error and, especially at higher levels of ac
tivity, counting error may constitute a gross understatement of the to
tal error that should reasonably be attributed to the measurement. Thi
s paper compares the measurement confidence limits obtained using a ty
pical approximate error propagation procedure with the ''exact'' confi
dence limits. A Monte Carlo error propagation method is also considere
d. The results of the three methods are compared using Rn-222 measurem
ent procedures for illustrative purposes.