Background and aims – Although the effectiveness of protected areas for the conservation of biodiversity is widely accepted, only very little direct scientific evidence exists. For small reserves embedded in hostile agricultural matrix this empirical proof is even lacking all together, although effectiveness can be most questioned here. In this study, we compared the plant species diversity of ten small nature reserves in Flanders, Belgium with the plant species diversity of the adjacent, agricultural landscape matrix. Methods – Plant species diversity was studied for ten nature reserves in Haspengouw, Belgium using 10 × 10 m ² survey plots at a density of one plot/ha. The same sampling density was applied to the direct vicinity of the reserve in the landscape matrix consisting of intensively managed agricultural landscape. Key results – An overall clear pattern of higher plant species diversity was found for nature reserves compared to landscape matrix at the plot, reserve and regional scale, clearly indicating the effectiveness of nature reserves. This pattern, however, appeared to be influenced by the soil type, with reserves on loamy soils containing higher diversity compared to reserves on sandy loam soils. Also species composition differed between reserves and the landscape matrix, with more threatened, less ruderal species and more stress tolerating species of moist, nitrogen poor soils occurring inside the reserves. Conclusions – These results provide a clear evidence of the effectiveness of nature reserves in the conservation of plant diversity. Reserves may be considered as having a conservation credit when compared to the agricultural landscape.