of an intensive use of broad-spectrum herbicides in comparison with one application of conventional pre-emergence herbicide. Numbers of arthropods were measured by three techniques: visual counts on plants, catches in pitfall and yellow sticky traps. Weed density was much higher in conventional treatment in the first year, showed significant difference in the second year, but was no significant difference in the third year. Counts of arthropod taxa were significantly different only in the first year in the two kinds of weed management systems. In visual counts Cicadellidae and Aphididae among herbivores, the two main generalist predators, Orius spp. and Araneae, and the family Coccinellidae were more abundant on plants treated twice with glyphosate. In pitfall there were higher records in glyphosate-treated plots for Myriapoda but the opposite was seen for Carabidae counts. The yellow sticky traps catches were higher in the glyphosate-treated plots for Cicadellidae and Mymaridae, and lower for Thysanoptera. Most of the significant differences found between herbicide regimes disappeared when abundances of weeds (monocotyledons and dicotyledons) were introduced into the analysis as covariates; this finding signals weed abundance as the main cause of arthropod abundance alteration. However, only a drastic alteration of weed abundance causes significant changes in arthropod densities.