Previous animal studies support the presence of an upper airway reflex mechanism that when blocked by topical anesthesia of the upper airway results in upper airway occlusion. We sought a similar reflex mechanism in humans. Nine normal male volunteers 20 to 28 yr of age underwent 3 successive overnight sleep studies: a control study (C); a study in which selective topical oropharyngeal anesthesia (OPA) was achieved prior to sleep using a 10% lidocaine spray and 0.25% bupivocaine solution; a study in which selective nasal anesthesia (NA) was achieved prior to sleep using a mixture of 2% lidocaine and 0.25% bupivocaine solutions instilled into the nose while the nasal airway was positioned as the most dependent part of the upper airway. Total sleep times were similar during the 3 study nights as were the amounts of slow-wave and rapid-eye-movement (REM) sleep. Obstructive apneas and hypopneas (OAH) differed significantly between the 3 study nights [13(3.8), mean (SEM), during OPA as compared to 3(1.8) during C and 7(2.5) during NA; p less than 0.01 by ANOVA] and were most frequent during REM sleep. Total apneas and hypopneas also differed significantly between the 3 study nights [19(3.9) during OPA as compared to 8(2.1) during C and 14(3.9) during NA; p less than 0.01 by ANOVA]. Movement arousals terminating periods of abnormal respiration also differed significantly [21(6.1) during OPA as compared to 12(3.6) during C and 14(4.6) during NA; p less than 0.05 by ANOVA]. No subject, however, developed clinically significant sleep apnea or significant oxygen desaturation during sleep.(ABSTRACT TRUNCATED AT 250 WORDS)