Cancer pain is a relatively neglected public health issue in Taiwan. To characterize the nature of this problem, interviews were conducted on newly diagnosed cancer patients admitted to the Tri-Service General Hospital during a period of 18 months. Data were collected on the prevalence and severity of cancer pain, its treatment, and impact on patients in the week before the interview. Correlates of prevalence and severity of cancer pain were also examined. The final analysis included 296 patients who had no history of recent surgery. Most of the patients (69%) were interviewed within 14 days of their definitive diagnosis of cancer. Thirty-eight percent (N = 113) of the patients had cancer-related pain. Of these 113 patients, 65% had "significant worst pain" (worst pain level at or above five on a ten-point scale) and 31% had "significant average pain" (average pain level at or above five most of the time); 69% received no pain medication at all or inadequate medication (not "by the ladder"), and 23% had pain medication that was not administered at a fixed interval (not "by the clock"). Multivariate analyses showed that cancer pain was more prevalent in non-Mainlanders, those with a lower level of insurance, those with a history of excellent pain tolerance, those with poor Eastern Cooperative Oncology Group (ECOG) performance status, and those with distant metastases. Patients who were at greater risk of "significant worst pain" were those with regional or distant metastases, those in whom an inadequate analgesic medication had been prescribed (not "by the ladder"), and those who had received an appropriate analgesic medication but no fixed schedule dosing ("by the ladder" but not "by the clock"). Patients who were at greater risk of "significant average pain" were those not undergoing any resection of the tumor lesion and those who received an appropriate drug but no fixed schedule dosing ("by the ladder" but not "by the clock").