Thirty eight specialists in one district health authority were asked to take part in a questionnaire survey to assess the appropriateness of referral and the quality of the referral letter for 20 consecutive new patients each. A total of 705 new patient referrals to 13 specialties were included in the study. Twelve of the 38 specialists were randomly selected and their 234 new patient referral letters were independently assessed by a general practitioner for the appropriateness of the referral decision. The study revealed errors and omissions in between 5% and 28% of referral letters according to the category of information. Thirteen per cent of the new patient referrals were assessed by specialists to be inappropriate and 4% of patients had been referred to an inappropriate specialty. Significantly more of the referrals to medical specialties were inappropriate (20%) than to surgical specialties (9%) (P < 0.01). There were more than three times the number of errors and omissions in the referral letters of referrals assessed as inappropriate than in the referral letters of referrals assessed as appropriate (P < 0.01). The referral letters of referrals assessed as inappropriate were more than nine times as likely to omit the reasons for or objectives of the referral compared with letters for those referrals assessed as appropriate (P < 0.01). There was a good overall agreement between the specialists and general practitioner in their assessment of the appropriateness of the clinical referrals (kappa = 0.614, P < 0.001).(ABSTRACT TRUNCATED AT 250 WORDS)