The correct implementation of the Durbin and Koopman simulation smoother is explained. A possible misunderstanding is pointed out and clarified for both the basic state space model and for its extension that allows time-varying intercepts (mean adjustments).