Forest transition, a concept introduced in the early 1990s by Alexander Mather, proposes a reversal of forest cover change trends: from long‐term deforestation to stable or increasing forest cover. Several well‐documented examples of forest transition have been described since its proposal, typically at the national level. Initially, forest transition was explained by endogenous drivers, yet recently researchers have linked forest transition to spatial interactions taking place over increasingly larger distances and land use displacements occurring beyond national boundaries. Therefore, in the telecoupled global land use system, country‐level analysis of forest cover changes includes several pitfalls highlighted by the Modifiable Areal Unit Problem (MAUP) and it may therefore not be appropriate to draw conclusions about the occurrence of forest transitions. In this study, we attempt to explain this problem with a simple land use change simulation that extends the model introduced by Mather and Needle in 1998, and case study data from the Carpathian Mountains in Poland. Our results suggest that an abrupt increase of interaction distance is sufficient to cause land use displacements beyond the boundaries of areal units, resulting in reversals of previous land use trends and forest transitions. With interaction distance exceeding the dimensions of areal units considered in the analysis, these areal units may be considered arbitrary, and forest transitions occurring within their boundaries spurious in the sense of MAUP. We conclude that this finding may apply to other contemporary country‐level analyses of forest cover change, and argue that at present, only the global‐level forest transition is MAUP‐insensitive, possessing properties relevant to its early theoretical meaning.