Hi Laura,
I'm sorry to hear you still have problems! I've discussed similar issues here. If the file is big as you report (around 40MB), it comes to no surprise: our system is not able to cope with such a large 'one off' import. Currently, the suggestion is to limit the import size to 2000-2500 references at the time. There is no need to re-do your PubMed search, though: you can just subdivide the text file into smaller chunks, making sure that you don't break references apart. If you know how many references are in your current file, you should be able to subdivide your imports in such a way that each file contains roughly 2000 references...
I am not pleased to impose such limits on our users, but so far I've been unable to find a viable workaround. On the other hand, searching PubMed from inside EPPI-Reviewer should allow you to import big searches in chunks (note the "import results from Item [...] to Item [...]" boxes at the bottom of the screen). Also, in case you are combining different PubMed searches into one, it might be a better idea to import each one of them individually: this will make the importing phase less troublesome (but not faster!) and will then give you more information about where single items are coming from. Note also that when searching PubMed from EPPI-Reviewer itself, the search string is automatically saved along with the source information: depending of the kind of work you are doing, this might turn out to be useful information later on.
I hope this helps,
Sergio