You may have close relatives who should appear but are not in your match list even though the relationship is correct.
I have an example with two kits B (kitno 306658) and D (kitno 324479). B is a triplet child of D (2C 1R). See my blog post from 2014 To secure my descent from Höijen.
When I did the Y37 test on them I thought it would match but it didn't. They do not appear in each other's match lists; they have genetic distance = 7 but the limit to be seen is four. Now I have upgraded to Y111 and then it will be genetic distance = 8 and since the limit is now ten, they are visible to each other. Only one distinguishing marker has been added on marker 38 to 111.
This is what the Y37 markers look like for B (top) and D:
Six markers separate and according to FamilyTreeDNA's way of counting it will be genetic distance = 7.
This is how the remaining markers look up to Y111:
Of these 74 markers, only one differs. This is a perfectly OK result, it is rather that there are unusually few deviations on these 74 markers. Since there is really no reason to doubt the relationship, I now feel quite convinced that it is indeed true according to the books.
The conclusion is that one should be careful about completely dismissing consanguinity just because it is not in the Y37 match list. It's always about chance and STR markers are capricious.
Jakob Norstedt
English (UK) 
Svenska