Rats were trained to discriminate between four dissimilar drugs (phenobarbital, nicotine, fentanyl, and methylphenidate) in compartments which contained either four identical levers or four dissimilar response manipulanda. During successive training sessions, the drug condition was cycled through the four training conditions. The objective was to compare speed of acquisition and asymptotic accuracy of discriminative control in the following types of compartments: (1) Undecorated compartments with four identical levers; (2) Compartments with four dissimilar response manipulanda (lever, wheel, nosepoke, panel); (3) Four-lever compartments with a unique sensory environment surrounding each lever; (4) Compartments with four dissimilar manipulanda, each surrounded by a unique sensory environment. The required four-drug discrimination were learned in all training compartments. Independent variables that produced statistically significant effects on speed of acquisition and/or asymptotic accuracy included drug, dosage, use/nonuse of four dissimilar response manipulanda, and presence/absence of environmental decorations around each manipulandum. Although the use of four different response manipulanda and/or the use of distinctive decorations surrounding each of the four manipulanda did increase speed of acquisition, these manipulations also resulted in biases towards/against particular individual environments or manipulanda during the acquisition phase of the experiment. Such biases can complicate the interpretation of results of conventional drug discrimination studies, especially if they persist into the asymptotic accuracy phase, which was not observed in the present study.