A procedure for combining human errors in structural risk analysis is presented. The procedure accounts for the fact that human errors cause unconservative and/or conservative changes in the probability density functions of the basic variables (i.e. resistance and load effect) involved in the structural risk model. The paper also includes results with regard to the sensitivity of the reliability index to changes in (a) the type of individual errors, (b) the type of combination of errors, and (c) the probability of occurrence of such errors. The results could be used to evaluate the significance of various combinations of errors on structural risk.