After head injuries, particularly mild or moderate ones, a range of post-concussion symptoms (PCS) are often reported by patients. Such symptoms may significantly affect patients' psychosocial functioning. To date, no measure of the severity of PCS has been developed. This study presents the Rivermead Post Concussion Symptoms Questionnaire (RPQ) as such a measure, derived from published material, and investigates its reliability. The RPQ's reliability was investigated under two experimental conditions. Study 1 examined its test-retest reliability when used as a self-report questionnaire at 7-10 days after injury. Forty-one head-injured patients completed an RPQ at 7-10 days following their head injury and again approximately 24 h later. Study 2 examined the questionnaire's inter-rater reliability when used as a measure administered by two separate investigators. Forty-six head-injured patients had an RPQ administered by an investigator at 6 months after injury. A second investigator readministered the questionnaire approximately 7 days later. Spearman rank correlation coefficients were calculated for ratings on the total symptom scores, and for individual items. High reliability was found for the total PCS scores under both experimental conditions (Rs = + 0.91 in study 1 and Rs = + 0.87 in study 2). Good reliability was also found for individual PCS items generally, although with some variation between different symptoms. The results are discussed in relation to the major difficulties involved when looking for appropriate experimental criteria against which measures of PCS can be validated.