We examined to what extent temporal dynamics of Jacobaea vulgaris cover in old-fields were related to plant–soil feedback, soil nutrients, seed availability and performance, and seedling establishment. Long-term measurements at an experimental field and in ten old-fields representing a chronosequence following land abandonment revealed a remarkably similar hump-shaped temporal pattern of J. vulgaris cover, which peaked at about five years after abandonment. In a plant–soil feedback study, J. vulgaris biomass of plants grown in soil from all chronosequence fields was lower than in sterilized control soil. However, biomass of J. vulgaris in the feedback study was lower when grown in soil collected from fields with a high density of J. vulgaris plants than in soil from fields with a low density of J. vulgaris. When plants were grown again in the conditioned soil, a strong negative plant–soil feedback response was observed for soils from all fields. These results indicate that soils from all stages of the chronosequence can develop a strong negative soil feedback to J. vulgaris, and that there is a positive relationship between J. vulgaris density and the subsequent level of control by the soil community. In a common-garden experiment with turfs collected from the chronosequence fields in which J. vulgaris was seeded, seedling establishment was significantly lower in turfs from older than from young fields. In a seed bank study the number of emerging seedlings declined with time since abandonment of the field. In conclusion, negative plant–soil feedback is an important factor explaining the hump-shaped population development of J. vulgaris. However, it is not operating alone, as propagule availability and characteristics, and competition may also be important. Thus, in order to explain its contribution to plant population dynamics, the role of biotic plant–soil interactions, soil nutrients and life history characteristics along successional gradients should be considered from a community perspective.