Chronic low back pain is both prevalent and costly in many industrialized nations. Although many modalities exist for the treatment of this condition, few are as commonly used or as controversial as the use of opioids. Many sets of guidelines exist for the prescription of opioids for chronic nonmalignant pain, but little evidence addresses what factors actually contribute to the decision to initiate and maintain patients on these drugs. In these studies we first identified 2 groups of 100 patients each, all with chronic low back pain. Group N patients received long-term nonsteroidal anti-inflammatory drug therapy for the treatment of their pain, whereas Group O received opioids long-term. The identities of the specific analgesics were tabulated. A list of variables including patient characteristics, healthcare utilization factors, and psychologic characteristics were extracted from their medical records. Regression analysis was performed, which resulted in the identification of 4 variables of age, depression, personality disorder, and history of substance abuse as being closely linked to the use of opioids for the treatment of back pain in preference to nonsteroidal anti-inflammatory drugs alone. By using the derived regression equation, 79% of patients could be correctly classified into Group O or Group N. Pain intensity did not predict opioid use. We present alternative explanations for these observations.