No matter what you draw there’s a similar logo. Especially in niche fields. Find ways to use this to your advantage.

Doing proper research about your client's competitive landscape is one of the most helpful things you can do.

Unites States:
http://www.uspto.gov/main/sitesearch.htm
Canada:
http://www.ic.gc.ca/app/opic-cipo/trdmrks/srch/tmSrch.do?lang=eng

It may be hard if your design can't easily be described in search terms of words and letters, but the information is there to be found.

I came across this article not too long ago that you might find interesting. Looking at the examples of similar logos, you can see how some are too close for comfort, and some aren't. For example, I think the Sun Microsystems diamond very clearly has the word "Sun" in it, and the Columbia Sportswear is doing its own thing. To me, they are different enough. But someone is always going to think your logo is similar to someone else's, so why worry about it? As long as you aren't actually ripping someone off (especially these guys), and have good reasons for the design choices you make, you should feel confident that your logo is unique enough.

Although Jackson does make an excellent point about checking out the competition. If the Sun/Columbia companies were in the same industry, THEN I might think twice about their logos.

I have a vague memory of a site filled with well-known logos, but I can't find it now. Perhaps it was just something I dreamed up.

There are plenty of books that periodically collect logos, too -- another resource for doing the research Jackson mentioned.

A logo that manages to capture some of the essence of a company/organisation can’t be wrong, similar or not. Since this is so very, very hard to achieve, stumbling upon a look-alike may evoke the notion that your own design is too random.

And yes, researching the clients competitive landscape is mandatory. Doing it too early in the process can obstruct the basic approach, though. Better to throw out all your favorite sketches later.