Although reflexes are recognized as protective responses to noxious stimuli, less is known about voluntary behavioral responses to cancer pain, which could provide clinicians with important diagnostic and therapeutic information. Forty-five patients with lung cancer were studied in their homes on 2 occasions to identify pain behaviors and to examine relationships between behaviors and selected variables. Patients completed the McGill Pain Questionnaire (MPQ) and Visual Analogue scale (VAS). Using a videotape observation method, patients sat, stood, walked, and reclined for 10 min. Videotapes were scored using 5 position-related and 31 pain-related behavior definitions. Within 3 days scored behaviors were described to patients who reported whether each scored behavior was performed: to express pain; because pain prevented usual behavior; to control pain; or as a habit. Patients reported that pain was controlled by 42 different behaviors; the number of different pain-reduction behaviors was correlated with pain intensity (r = 0.44) and pain quality (r = 0.64). Simultaneous multiple regression indicated that length of time pain was experienced, number of pain sites, pain quality, and pain intensity accounted for 41% of the variance in the number of pain control behaviors. None of the taped behaviors was reported as performed to express pain, and few of the patients reported that pain prevented behavior during the video session. Results clarify the pain-behavior construct, provide insight about the multidimensional nature of lung cancer pain, and suggest directions for behavioral interventions to augment pharmacological therapy for lung cancer pain.