The same authors give conflicting recommendations in different articles. It's clear this is still an open issue and, I expect, conservative docs will go for CYA.
-------
: Curr Pain Headache Rep. 2008 Apr;12(2):128-31.
Cluster headache: to scan or not to scan.

Favier I, Haan J, Ferrari MD.

Department of Neurology, K5-Q, Leiden University Medical Centre, PO Box 9600, 2300 RC Leiden, The Netherlands. i.favier@lumc.nl

By definition, cluster headache (CH) is not caused by underlying structural pathology. However, patients with CH or CH-like syndromes and an associated structural lesion have been described. In many cases it is difficult to establish a causal relation between the headache syndrome and the lesion. We reviewed the literature for symptomatic CH or CH-like cases in which causality was very likely, and we found that even typical CH with a typical episodic time pattern and a response to typical CH treatment can be caused by underlying structural pathology such as a pituitary tumor. Based on this small retrospective series of case reports, it is impossible to give advice about neuroimaging.

If neuroimaging is considered, MRI (not CT) is the investigation of choice.

PMID: 18474193 [PubMed]

Hmm IMO a scan always should be done, even if headaches look exactly like CH. MRI or at least CT... (though very often some things are way too small to be visible @ CT image)

Well, I've got a pituitary tumor (microadenoma) but it doesn't seem to be related with my CH-like headaches. I was also told (by a very good doctor) that some patients had their tumors (bigger than mine) removed and it didn't help them... My tumor was very small when it was detected, I am very curious how is it at the moment (I am having a MRI next week).