Apart from taking up server space pages that are no longer maintained or of an older layout they will continue to be spidered. This is an unnecessary load on both your server and a waist of time for the robot.
If the page is removed from the server it will generate a 404. In this case this is a good idea as the page will eventually be removed from the indices of search engines.
In my SEO Experiments, namely the Craft Fair Project, I started to monitor server logs. I noticed quite a lot of spider activity retrieving pages that I no longer wanted to be in any index. I started to remove the ones that I didn't want.
This page was not uploaded to the server and therefore was not indexed. This shows the merits of doing an on-line audit of my pages. This is an area where developing pages off-line is somewhat problematic.
In my upgrade cycle of layut design there are a number of pages that were replaced with pages with _3 suffix on the file name. This turned out to be a bad idea as I then had two versions on the server. I have since formatted old layout(ed) pages with the same file name, but I now have a mixture of page with _3 and others without and no way of determining whether that are both required and/or converted to the latest layout.