This paper addresses one aspect of the calibration of a 109Cd based K-shell x-ray fluorescence measurement system, namely the treatment of the calibration line intercept. Under circumstances of contamination, the intercept may warrant statistical treatment different from that currently performed. This paper proposes refinements to the existing method of subtracting the phantom calibration line intercept from in vivo responses in the calculation of in vivo concentrations. These refinements are recommended because the existing method can underestimate in vivo concentrations by a small amount under normal operating conditions. Contamination of the calibration standard matrix of plaster of Paris by both lead and non-lead contaminants is addressed.