Neurology manuals generally recommend odor identification for simple assessment of olfaction. Nevertheless, even patients with normal olfaction (normosmics) often perform only poorly. Three experiments demonstrate that such an ambiguous outcome will disappear if the test incorporates highly familiar substances and, more important, a procedure to circumvent the olfactory-verbal gap that frequently separates an odor from its name. One multiple-choice procedure, for instance, led to 100% accuracy among normosmics. Another led to 99% accuracy among normosmics and 0% accuracy among anosmics. The investigation also reveals that scratch-and-sniff labels could possibly replace customary odorants in the clinical test.