The optimal assessment of cancer pain includes a detailed description of pain characteristics and classification by both syndrome and likely mechanisms. In the clinical setting, the interpretation of this information is aided by knowledge of the available clinical experiences on these aspects of the pain. Unfortunately, existing data are limited. There have been few large surveys of cancer pain characteristics and syndromes, and comparative data from patients in different parts of the world are entirely lacking. To better define the characteristics of cancer pain syndromes the Task Force on Cancer Pain of the International Association for the Study of Pain (IASP) conducted a prospective, cross-sectional, international, multicenter survey of pain specialists and their patients. From a total of 100 clinicians who described themselves as cancer pain practitioners in the IASP membership directory, 51 agreed to participate in the survey and a total of 58 provided data. These clinicians resided in 24 countries and evaluated a total of 1095 patients with severe cancer pain mostly requiring opioid medication, using a combination of patient-rated and observer-rated measures. The patient-rated scales comprised a pain intensity measure chosen from the brief pain inventory. The observer-rated information included demographic and tumor-related data, and responses on checklists of pain syndromes and pathophysiologies. Patients were heterogeneous in terms of demographics and tumor-related information. More than 76% had a Kamofsky performance status score < or = 70. Almost one-quarter of the patients experienced two or more pains. A large majority of the patients (92.5%) had one or more pains caused directly by the cancer; 20.8% of patients had one or more pains caused by cancer therapies. The average (SD) duration of pain was 5.9 (10.5) months. Approximately two-thirds of patients (66.7%) reported that the worst pain intensity during the day prior to the survey was > or = 7 on a 10-point numeric scale. The factors that were univariately associated with higher pain intensity included the presence of breakthrough pain, somatic pain or neuropathic pain, age younger than 60 years, and lower performance status score. A multivariate model suggested that the presence of breakthrough pain, somatic pain, and lower performance status were the most important predictors of intense pain. Pains that were inferred by the treating clinician to be nociceptive and due to somatic injury occurred in 71.6% of the patients. Pains labeled nociceptive visceral were noted in 34.7% and pains inferred to have neuropathic mechanisms occurred in 39.7%. In a broad classification, the major pain syndromes comprised bone or joint lesions (41.7% of patients), visceral lesions (28.1%), soft tissue infiltration (28.3%), and peripheral nerve injuries (27.8%). Twenty-two types of pain syndromes were most prevalent. Large differences in the diagnosis of breakthrough pain by clinicians of different countries suggest that this phenomenon is either defined or recognized differently across countries. These data confirm, in segment of the cancer population experiencing severe pain, in different parts of the world, that cancer pain characteristics, syndromes and pathophysiologies are very heterogeneous. Predictors of worsening pain can be identified. The data provide a useful context for the interpretation of pain-related information acquired in both clinical and research settings. They suggest the need for future studies and the potential usefulness of a written checklist for cancer pain syndromes and pathophysiologies.