Two cow traffic situations were tested sequentially in an automatic milking system (AMS) for effects on cow behavior, effective use of the barn, and milking capacity. The first situation was forced cow traffic: 63 cows had to pass through the AMS to go from the lying area to the feeding area. The second was semiforced cow traffic: 67 cows (60 cows from before) had free access to a forage feeding area at one end of the barn but could only access an area with concentrate feeders by passing through the AMS. Behavior of all cows was monitored as well as for two subsets of cows present in both situations: 8 cows with low frequency vs. 7 cows with high frequency of visits to the AMS. In each situation, cows were observed for 72 h. Cow locations and behavior were noted at 10-min intervals for all cows and individually for the selected cows. In semiforced traffic, the herd readily used the freely accessible forage feeding area, ate longer (17.4% of the day vs. 15.1 ± 0.59%), and stood less in freestalls (9.0 vs. 11.8 ± 0.30%) than when cow traffic was forced. Nonmilking visits to the AMS tended to decrease, whereas milking visits remained unchanged in the semiforced situation. The subset of cows that visited the AMS more often had fewer nonmilking visits (1.8 vs. 4.2 ± 0.7) in the semiforced traffic situation whereas cows with low frequency of visits to the AMS had a nonsignificant increase (1.5 vs. 1.0) in nonmilking visits. Cows that visited the AMS frequently used the forage feeding area and the lying area next to it more than low frequency cows and use of those areas increased further during the semiforced situation. The semiforced cow traffic was deemed more desirable than forced cow traffic both for cows and for the capacity of the automatic milking system.