Cluster headache is a primary headache by definition not caused by any known underlying structural pathology. Symptomatic cases have been described, for example tumours, dissections and infections, but a causal relationship between the underlying lesion and the headache is difficult to determine in many cases. The proper diagnostic evaluation of cluster headache is an issue unresolved. The literature has been reviewed for symptomatic cluster headache or cluster headache-like cases in which causality was likely. The review also attempted to identify clinical predictors of underlying lesions in order to formulate guidelines for neuroimaging. Sixty-three cluster headache or "cluster headache-like"/"cluster-like headache" cases in the literature were identified which were associated with an underlying lesion. A majority of the cases had a non-typical presentation that is atypical symptomatology and abnormal examination (including Horner’s syndrome). A striking finding in this appraisal was that a significant proportion of CH cases were secondary to diseases of the pituitary gland or pituitary region. Another notable finding was that a proportion of cluster headache cases were associated with arterial dissection. Even typical cluster headaches can be caused by structural lesions and the response to typical cluster headache treatments does not exclude a secondary form. It is difficult to draw definitive conclusions from this retrospective review of case reports especially considering the size of the material. However, based on this review, I suggest that neuroimaging, preferably contrast-enhanced magnetic resonance imaging/magnetic resonance angiography should be undertaken in patients with atypical symptomatology, late onset, abnormal examination (including Horner’s syndrome), or those resistant to the appropriate medical treatment. The decision to perform magnetic resonance imaging in cases of typical cluster headache remains a matter of medical art.