Many tests of olfactory dysfunction are either too complex, too expensive, or too time-consuming to be of use in routine clinical testing. Thus, the present multicenter study was undertaken to investigate a new approach, the so-called "random" test. In this test different concentrations of citronellal and phenyl ethyl alcohol are applied according to a pre-established order; patients are asked to identify the odor if possible. The test score is the sum of correctly identified odors. Test administration takes about 10 min. Two studies were performed. Basic characteristics of the test were explored in experiment 1 in 176 healthy subjects (76 male, 100 female; age 12-85 years, mean age 30 years), namely test-retest reliability, correlation with other measures of olfactory sensitivity, and sensitivity of the test to differences in age and gender. In the second experiment the test was tried in 97 patients (45 male, 52 female; age 19-78 years, mean age 47 years) in a clinical environment to investigate its usefulness in diagnosing olfactory loss. The "random"-test was found (1) to exhibit a test-retest reliability similar to that reported for established measures of olfactory function (r = 0.71; P < 0.001), (2) to correlate with other measures of olfactory sensitivity (0.82 > r > 0.60; P < 0.001), (3) to differentiate between expected differences in olfactory sensitivity in relation to gender (t > 2.602, P < 0.011), and (4) to discriminate between different degrees of olfactory loss (F > 36.6, P < 0.001). Based on these data, and the fact that the new test requires little time and is easy to use, this approach can be expected to suit clinical needs.