BACKGROUND: Headache disorders are under-recognized and under-diagnosed. A principal factor in their suboptimal management is lack of headache-related training among health-care providers, especially in primary care. In Estonia, general practitioners (GPs) refer many headache patients to neurological specialist services, mostly unnecessarily. GPs request "diagnostic" investigations, which are usually unhelpful and therefore wasteful. GP-made headache diagnoses are often arcane and non-specific, and treatments based on these are inappropriate. The aim of this study was to develop, implement and test an educational model intended to improve headache-related primary health care in Estonia. METHODS: This was a controlled study consisting of baseline observation, intervention and follow-up observation using the same measures of effect. It involved six GPs in Põlva and the surrounding region in Southern Estonia, together with their future patients presenting consecutively with headache as their main complaint, all with their consent. The primary outcome measure was referral rate (RR) to neurological specialist services. Secondary measures included number of GP-requested investigations, GP-made headache diagnoses and how these conformed to standard terminology (ICD-10), and GP-recommended or initiated treatments. RESULTS: RR at baseline (n = 490) was 39.5 %, falling to 34.7 % in the post-intervention group (n = 295) (overall reduction 4.8 %; p = 0.21). In the large subgroup of patients (88 %) for whom GPs made clearly headache-related ICD-10 diagnoses, RR fell by one fifth (from 40 to 32 %; p = 0.08), but the only diagnosis-related RR that showed a statistically significant reduction was (pericranial) myalgia (19 to 3 %; p = 0.03). There was a significant increase towards use of more specific diagnoses. Use of investigations in diagnosing headache reduced from 26 to 4 % (p < 0.0001). Initiation of treatment by GPs increased from 58 to 81 % (p < 0.0001). CONCLUSIONS: These were modest changes in GPs' entrenched behaviour. Nevertheless they were empirical evidence that GPs' practice in the field of headache could be improved by structured education. Furthermore, the changes were likely to be cost-saving. To our knowledge this study is the first to produce such evidence.