After split-testing two ads, do you consider any factors besides CTR before deciding to kill one of the ads? Of course, you want to make sure the serve ratio is close to 50/50 and that both ads have accrued a lot of clicks. But besides that...?
CPC, too. (If you go to the Ad Variations tab and click the "Create an ad variation" link, it'll then show you at the bottom a summary of the CTR and CPC of each active ad.)
Frequently, improving your clickthrough rate will cause your conversion rate to fall, if you aren't careful. If you are trying to get as many people as possible to click on your advert, being specific about what you offer works against you. For example "Get Free Information Here" will get more clicks than "Sign Up Here For Free Info E-Mails" if you are trying to get subscribers to a newsletter. But the conversion rate will be less, as you aren't making your service clear in the advert. The real problem here is that you have to run a test far longer to see if it's affected the conversion rate than to see if it's improved the clickthrough rate... I've written a blog on this (see my blog link) going into more detail, entitled "A Cautionary Tale..."
Consider also whether low performing ads would receive a better CTR if they were in a different adgroup. For instance, you may have an ad that's getting a CTR of only 2% for an adgroup targeted to the keywords, "buy widgets". But the same ad may achieve a 10% clickthrough in a different ad group targeted to the phrase "where to buy durable widgets". There's plenty of other factors to consider.
Focus on conversions. If both ads have equal conversion rates (or similar conversion rates), eliminate the ad with the lower CTR. CustardMite makes good points about improving CTR too much and how this can lower conversions. Always test to see what works best for you.