Uncertainty included in forest variables is normally ignored in forest management planning. When the uncertainty is accounted for, it is typically assumed to be independently distributed for the criteria measurements of different alternatives. In forest management planning, the factors introducing the uncertainty can be classified into three main sources: the errors in the basic forestry data, the uncertainty of the (relative) future prices of timber, and the uncertainty in predicting the forest development. Due to the nature of these error sources, most of the involved uncertainties can be assumed to be positively correlated across the alternative management plans and/or criteria. This, in turn, may lead to overestimating the risks due to the uncertainty. In this study, we show how the SMAA-2 method can be employed for dealing with dependent uncertainties in strategic forest planning. The uncertainties in the criteria measurements are assumed to follow a multivariate normal distribution. The correlations between the variables are assessed based on expert judgment. A case problem is analysed both without and with dependency information to illustrate the practical importance of the dependencies. The results show that ignoring the dependencies will markedly weaken the support for the decisions to be made.