This paper concerns the hypothesis that shoot (light) competition is the main interaction determining the community change during succession from a relatively species-poor deciduous forest (an overgrown former grassland) to a species-rich grassland, while root (nutrient) competition is of little importance. In a 4-yr restoration experiment, clearcutting, mowing and root trenching were used as treatments. The results did not reject the hypothesis. However, the significance of separating two kinds of shoot competition - ‘coarse-scale (between different growth forms) and ‘fine-scale’ (between similar growth forms) became evident. Release from the ‘coarse-scale’ shoot competition (between different growth forms) increased species richness at the beginning of the experiment. This change was interpreted as the replacement of one species pool (shade-tolerant herbaceous perennials) by another (light-demanding herbaceous perennials), the second pool containing considerably more species. The importance of ‘fine-scale’ shoot competition increased gradually - the levelling of competition by mowing resulted in a more pronounced increase in species richness during successive years. The elimination of ‘coarse-scale’ root competition seemed to be important to some extent only in combination with another treatment - mowing. Initial colonization of the cleared area by individual species was a stochastic process which had little relevance to life-history traits. True grassland species were able to colonize quickly. On the community scale, the developing community still remained relatively poor in species. In all plots which were cleared but not mown, succession already started to reverse towards woody vegetation in the third year.