Diagnosing cause of pain in children with severe cognitive impairments is difficult due to their problems with communication. Identification of risk factors for specific pain etiologies might help professionals in this task. The aim of this study was to determine whether child-related characteristics increase risk for specific types of pain. Participants were the caregivers of 41 females and 53 males with moderate to profound mental retardation, who were aged 3 to 18 years 8 months (mean 10:1, SD 4:4) but who communicated at the level of a typical child of 13.8 months (SD 10 months): 44 of the children had cerebral palsy (CP) and 59 a seizure disorder. Caregivers reported the cause of children's episodes of pain for four 1-week periods over 1 year. Logistic regression analyses were used to predict occurrence of specific types of pain using children's demographic, medical, and physical characteristics. Children had 406 episodes of pain due to accident, gastrointestinal conditions, musculoskeletal problems, infection, recurrent conditions, and common childhood causes. Results indicated that a unique set of risk factors predicted each pain type in this sample. Significant risk factors for pain included: lack of visual impairment and leg impairment (accidental pain); seizures, leg impairment, and greater number of medications (non-accidental pain); being male and tube fed (musculoskeletal pain); age <7 years, absence of CP, visual impairment, and less frequent medical monitoring (infection pain); being female and with arm impairment (gastrointestinal pain); and being tube fed and taking fewer medications (common childhood pains). In most cases, models were more specific than sensitive, indicating that the significant predictors are more useful for eliminating potential pain causes. These results suggest that population risk factors may be helpful in structuring diagnostic investigations for individual children with severe cognitive impairments.