Soil surface contains huge amount of seed of different weed species. By examination of weed seed distribution in the soil, it can be predicted which weeds will occur under different crops and to what extent. Detailed data analysis enables determination of timely soil tillage, as well as herbicide use. In the period 2008-2010, the weed seed bank was studied in different localities on the territory of Vojvodina (Krivaja, Ratkovo, Zmajevo, Bac, and Despotovo). Soil samples were collected from fields under different crops (soybean, sugar beet, maize, clover and sunflower). The sampling was performed diagonally in four replications from several arable depth layers of 0-10, 10-20 and 20-30 cm. Each sample was sieved through the system of cooper sieves of different diameter, according to the method of Conn (1987). After extraction from the samples, weed seeds were determined (Skender et. al., 1998; Kronaveter and Boar, 1994) and abundance of weed seed per m 2 was established for all three studied soil layers. The most abundant weed species in the weed seed bank under all studied crops were: Amaranthus retroflexus, Chenopodium album, Datura stramonium, Polygonum lapathifolium and Solanum nigrum, while less abundant were seeds of: Echinochloa cruss-galli, Euphorbia helioscopia, Sinapis arvensis, Sorghum halepense and Stachys annua. The highest abundance of weed seed was found at locality Zmajevo under sunflower crop in the soil layer of 0-10 cm, in the amount of 8.487 seeds per m 2. At locality Krivaja, under soybean crop, the highest number of seeds was recorded in layers of 10-20 and 20-30, in the amount of 9.252 and 10.609 seeds per m 2, respectively. The results were statistically processed and it was calculated that the highest coefficient of variance, i.e. the maximum deviation between number of seeds of weed species per soil layers was 40.53% at locality Despotovo.
