The objective of this experiment was 1) to quantify drift between bumblebee, Bombus terrestris L., colonies when hives are vertically stacked; and 2) to measure the impact of drift on the colony growth. The experiment was conducted outside, in an open area, at three sites with one stack of three colonies per site. Observations were made from the 28th to the 58th d of the colony development, before the competition point. Bumblebees were regularly marked with a colony-specific color to access the migrations and the cumulative births in colonies. The population size and the colony composition were analyzed by dissection at the end of the experiment. The results showed that ≈50% of the individuals observed at a hive entrance (entering and exiting individuals) were foreigners. The cumulative births became significantly lower on the lowest floor after one week of experiment. At the end of the experiment, colonies were composed of 20.58% drifters (resident since at least 1 d), and a mean migration rate of 19.82% was determined. Globally, the lowest floor seemed to be progressively abandoned by resident workers for the benefit of the other floors. Because drift induces modifications of the colony growth (potentially on the number of foragers produced and likely on pollination), we suggest to greenhouse growers not to stack colonies but to distribute them throughout the greenhouse.