Laboratory lore cautions that the behavior of cultured organisms can depend on subtle environmental factors and may sometimes even differ across laboratories. This is despite considerable efforts to employ identical diets and other prescribed environmental conditions at every location. Here we aim to demonstrate that laboratory dependence is real and quantifiable and that it raises interesting and valuable questions of general biological importance. We show that a simple bacterial community exhibits strikingly different behavior between two laboratories, in spite of standardized laboratory growth protocols. The community is composed of two Escherichia coli genotypes in a serial culture (seasonal) habitat. Previously, the two strains exhibited negatively frequency-dependent fitness which promoted their ecological coexistence at near equal frequencies. We show that when the pair of strains is moved to another laboratory they retain frequency-dependent fitness, but a new equilibrium is established in favor of one competitor. We attribute the altered competitive behavior to an unexpected difference in the culture medium between labs, most likely in the composition of deionized water. We show theoretically that the altered equilibrium could result from changes in each of two important growth parameters, Vmax (maximal growth rate) and Ks (resource concentration where growth is half-maximal). However, our empirical measurements rule out a large contribution by the parameter Ks.a