The simulation of vector autoregressive (VAR) models has raised issues concerning the application and interpretation of shocks to endogenous and exogenous variables in VAR, and consequently linear dynamic simultaneous equation models, when innovations are correlated across equations. If shocks are to use sample information and be based on innovation covariance matrices, then the actual shocks applied will be calculated from estimates of these matrices, and hence subject to sampling variability. The objective of this paper is to present confidence regions for multipliers which recognize the sampling variability of the estimated shocks. The model and its multipliers are described in Section 2 and the estimation of the model together with the asymptotic distribution of the model estimates are set out in Section 3. The formulae for the multiplier asymptotic confidence regions are derived in Section 4 and the results are discussed in Section 5.