Eighty-five lactating Holstein dairy cows in loose housing conditions in 2 symmetrical pens, each containing 28 feeding places, 3 waterers, and 1 automatic milking system (AMS), were used to evaluate the effects of the traffic type imposed on lactating cows through an AMS on milking frequency, feeding behavior, and milk production. The study was a crossover design with 2 periods and 2 treatments. Each period lasted 3 mo, with 1 mo of adaptation within each period. All cows were fed a partial mixed ration twice daily and up to 3 kg/d of a concentrate during the visits to the AMS. Treatments consisted of allowing free traffic of cows throughout the pen or forcing cows to pass through the AMS to access the feed troughs (forced traffic). Individual eating behavior and feed consumption were continuously monitored throughout the study using a computerized system. Individual milk production was recorded at each milking, and milk composition was recorded monthly. In addition, the number of cows brought to the AMS was recorded. The number of daily meals was less, whereas meal duration and meal size were greater with forced traffic (6.6 +/- 0.3 meals/d, 20.4 +/- 0.65 min/meal, and 2.7 +/- 0.09 kg/meal, respectively) than with free traffic (10.1 +/- 0.3 meals/d, 15.7 +/- 0.65 min/meal, and 1.8 +/- 0.09 kg/meal, respectively). Total dry matter intake (21.1 +/- 0.5 and 20.4 +/- 0.58 kg/d, respectively) and milk production (29.8 +/- 0.79 and 30.9 +/- 0.79 kg/d, respectively) were similar in the 2 systems. The number of voluntary and total daily milkings was greater with forced traffic (2.4 +/- 0.04 and 2.5 +/- 0.06 milkings/d, respectively) than with free traffic (1.7 +/- 0.06 and 2.2 +/- 0.04 milkings/d, respectively). Forced traffic improved the number of voluntary milkings, but altered milk quality and eating behavior of dairy cattle.