<jats:p>Abstract. A strong temperature increase in the Arctic is expected to lead to latitudinal treeline shift. This tundra–taiga turnover would cause a positive vegetation–climate feedback due to albedo decrease. However, reliable estimates of tree migration rates are currently lacking due to the complex processes involved in forest establishment, which depend strongly on seed dispersal. We aim to fill this gap using LAVESI, an individual-based and spatially explicit Larix vegetation simulator. LAVESI was designed to simulate plots within homogeneous forests. Here, we improve the implementation of the seed dispersal function via field-based investigations. We inferred the effective seed dispersal distances of a typical open-forest stand on the southern Taymyr Peninsula (northern central Siberia) from genetic parentage analysis using eight nuclear microsatellite markers. The parentage analysis gives effective seed dispersal distances (median ∼10 m) close to the seed parents. A comparison between simulated and observed effective seed dispersal distances reveals an overestimation of recruits close to the releasing tree and a shorter dispersal distance generally. We thus adapted our model and used the newly parameterised version to simulate south-to-north transects; a slow-moving treeline front was revealed. The colonisation of the tundra areas was assisted by occasional long-distance seed dispersal events beyond the treeline area. The treeline (∼1 tree ha−1) advanced by ∼1.6 m yr−1, whereas the forest line (∼100 trees ha−1) advanced by only ∼0.6 m yr−1. We conclude that the treeline in northern central Siberia currently lags behind the current strong warming and will continue to lag in the near future. </jats:p>