Got a question recently about how the indexer copes with major and minor versions and which results are displayed when you actually receive a search result. Here are some details I found:
The design is that we index whichever version of the document the crawler has access to. If the crawler account only has "full read" access to the web application being crawled, users cannot search over content present in unpublished/draft versions of the document.
On the other hand, if the crawler account has "full control" access to the web application being crawled, users will be able to search over content present in unpublished/draft versions.
The easiest check is to open the item in the browser using the crawler account and check what it can see.
Some other remarks:
One thing to note is that when you test te behavior, please make sure that your last changes are reflected in the index, otherwise you might see inconsistant search results.