Standard tests of murine unconditioned anxiety such as the elevated plus-maze and light-dark test are based on a dichotomy of avoidance behaviour (walled vs. open arms and dark vs. light compartments). We combined the principles of both tests by modifying the elevated plus-maze as follows: one walled arm was made transparent and had a white floor (WTW), whereas the other walled arm was opaque-gray having a black floor (WOB). Furthermore, one open arm had a white floor (OW), while the other had a black one (OB). These modifications allow the distinction between more than two sub-compartments that elicit different degrees of avoidance behaviour, thus having a higher discriminative potency. Additionally, the paradigm was thought to permit the within-task detection of pharmacological side effects on the perception of the anxiogenic stimuli provided. The degree of avoidance of the sub-compartments exhibited by saline-treated mice for the distal parts of the four arms was distributed as follows: WOB<WTW=OB<OW. This pattern demonstrates that the sub-compartments elicited at least three differed degrees of fear in control mice, and that these were able to discriminate between bright/dark compartments WOB vs. WTW and OB vs. OW. Diazepam given at 1, 2, and 3 mg/kg doses increased the number of entries into the distal part of the most aversive open white arm and increased the total time spent on the undivided open white arm. Both the 2 and 3 mg/kg dose increased motor activity and impaired bright/dark discrimination for the open but not for the walled arms. Thus, the graded anxiety test might be useful to screen for substances that retain the normal perception of anxiogenic stimuli, but prevent the transact of fear into undue panic reactions.