In view of the answers given, I would like to approach the answer to this question
as did R' Herczeg (who translated for the Artscroll Rashi series) in his book "Patterns in Rashi",
in which he takes an approach similar to Josh's answer.
To Rashi, the line between drash and pshat is vague. He viewed pshat as the simple meaning of the pasuk,
but only insomuch as it adheres to it grammatically and contextually. Whenever Rashi could not find
a pshat that fit with the grammar of the pasuk, he looked to Chazal's collection of aggados. He felt
that if an explanation of chazal fits with the grammar and context of the pasuk, it falls into the category
of pshat. You might say he felt that it was "pshat enough", even though it might not be the simplest explanation.
Many times we will find Rashi bring the popular pshat explanation, then bring a medrash from chazal. This is
because he saw some sort of flaw or redundancy in the pasuk that needed to be explained. To Rashi, that is included
Besides for grammar, several other things caused Rashi to side with chazal, including juxtaposition of topics,
unusually spelled words, and seemingly unnecessary information given by the Torah. Even though these topics are generally
considered, even by Rashi, as needing only to be addressed with drash, and according to pshat they may be ignored,
still in certain instances Rashi felt that for some specific reason they interfered with pshat on that occasion,
and therefore he may, in such instances, use chazal's aggados to clear things up. He considered this to be
consistent with the pshat approach.
Other later rishonim disagreed with Rashi. Meforshim like Ramban, Ibn Ezra, and Rashbam viewed pshat and drash as
distinct categories, reserving the simplest explanation of the pasuk and that alone for pshat. What about the
grammatical inconsistensies in the pesukim? We will have to explain them away continuing to use pshat. There is no
need for the more fantastical explanations of chazal. Those are drash; they have another place, a separate place,
in the interpretation of the pasuk.
This is what is meant by the famous argument Rashbam records that he had with Rashi. Rashbam felt that even the
irregularities in the text can be addressed with pshat, without the need to resort to aggados chazal. He
claims that Rashi conceded to him and admitted that if he had had more time, he would have included in his commentary
the simpler explanations that were still consistent with the grammar of the pasuk, as they were "more pshat" than
For an extensive discussion on this topic, with an abundance of examples, I highly reccomend R' Herczeg's book.