Sounds like you're doing it right.  Your post seems a little bit ambiguous as to whether or not you really ran an actual gel to separate out uncut vector.  If you didn't, do it.  But if that is what you did, then it's hard to see how you could get carryover.  Perhaps a secondary structure, or some other wicked unknown science specific to your sequence is getting in the way.  

This is the ideal process.  I don't see what you could do better, short of changing the strategy completely.  If you clone often, it might be worth the investment to learn Gibson's Assembly or a similar in-vitro recombination method.  These kinds of problems really do largely go away if you have 20+ bp directing the specificity of the ligation, and 50C temperature melts all kinds of structures that would get in your way in traditional T4 ligation. 

Thank you for the advice John. Sorry for the ambiguity, I did indeed run my plasmid on a 0.8% TAE agarose gel to separate out any undigested form of the vector.  I have attached one of my gels, the double digested vector is on the left and my digested insert on the right.

I am now wondering about the purity of my plasmid midi prep, it appears to have a couple extra bands, more than the expected relaxed/nicked/supercoiled forms. However when digested it, only one band is present. I have attached a second gel with the undigested form on the right, however perhaps the smudging is from an electrophoresis issue as the digested form is also smeared.

Hi there,

You should run the usual ligation controls for comparison: digested vector alone and digested vector alone with ligase. The former will tell you about the possible presence of undigested plasmid and the latter about the selfligation event. As you get transformants and no insert, I would think about uncomplete digest of vector and/or insert which prevents the expected event to occur and therefore favors the unwanted events... How close are the restriction sites on the vector and how close to ends are the sites on the PCR product? These could affect digest efficiency. If sites are close on the vector the 2 enzymes might compete for DNA. Sequential digest may help to improve digest efficiency. In theory there is no need to dephosphorylate the vector when performing double digest.

Hi Dominique,

The digestion sites on the vector are 62 bp apart and I do see a small band well below 250 bp on my ladder that I presume is this part I am losing. On the insert I have approximately 5 nucleotides preceding the restriction site on each end. Do you think I require a larger end for the enzymes? 

Hi again,

Well there doesn't seem to be any problem about the location of cutting sites in vector and in the PCR product... The only other issue I think of might be the star activity of BamHI enzyme if using non HF or FD enzyme...