Almost no experimental analysis of blushing has been done since Darwin's observations in 1872. Forty-eight college women watched a videotape intended to elicit blushing, and a videotape not intended to elicit blushing, but elicit physiological responses. A subject was alone, or with one or four persons present. Blushing, which was measured directly with a photoplethysmograph probe on the cheek, was greater during the blushing than nonblushing stimulation. Blushing increased as audience size increased from one to four, but not from zero to one. Audience size and kind of stimulation interacted statistically. Similar results were obtained with ear coloration, cheek temperature, and skin conductance responses, although confidence levels were lower. Cheek coloration and temperature were significantly correlated during nonblushing stimulation, and the zero and one audience conditions, but not during the four audience condition, when blushing was greatest. These results may be placed within the context of emotional effects of audience size generally, including stuttering and speech disturbance, disruption of learning, and self-reported tension.