Three-year field experiments were carried out on soybean in northeastern Croatia from 2007 to 2009 to assess weed control efficacy for some herbicides and their effects on crop yields. The herbicides were used in preemergence and preplus post-emergence, single or split applications, at different rates and in different combinations. Main weeds were annual grass of Echinochloa crus-galli (13-75 shoots m-2), broad-leaved species of Ambrosia artemisiifolia (8-30 plants m-2) and Chenopodium album (3-19 plants m-2). Total weed density was higher in the wet season of 2008 by 173% and 89%, compared to the drier seasons of 2007 and 2009, respectively. Considering the pre-emergence application, herbicide combination with alfa-metolachlor (960 g a. i. ha-1) + metribuzine (350 g a. i. ha-1) + clomazone (168 g a. i. ha-1) gave 97% of total weed control in 2008 with adequate moisture before and after the application. In the unfavorable seasons of 2007 and 2009 the same treatment showed a lower but still good weed control (94%). Considering the pre-em. plus postemergence herbicide applications, (s) dimetenamid (1008 g a. i. ha-1) plus imazamox, as a split (32 g + 16 g a. i. ha-1) provided the best control of main weeds (at least 90%) and consistently effective total weed control (94 98%) in all years. The same herbicide combination with a lower rate of (s) dimetenamid (720 g a. i. ha-1) plus imazamox (40 g a. i. ha-1), as a single application showed less A. artemisiifolia and C. album control in 2007 (up to 83%) and its affected crop yield. A similar efficacy was achieved with imazamox at same rate and mode of application when alfa metolachlor (960 g a. i. ha-1) + metribuzine (350 g a. i. ha-1) were applied pre-emergently (92%), with no damage to crop yield. Across-year differences in crop yield were insignificant between the pre-em. and pre-em. plus post-emergence herbicide treatments, ranging from 3186 kg to 3422 kg ha-1 [ABSTRACT FROM AUTHOR]