The Library of America posted Robert Heinlein’s comments from a lecture series in 1957:

First let us decide what we mean by the term “science fiction” — or at least what we will mean by it here. Anyone wishing a scholarly discussion of the etymology of the term will find one by Sam Moskowitz in the February, 1957 issue of The Magazine of Fantasy and Science Fiction. I shan’t repeat what he has said so well but will summarize for our immediate purposes. The field has existed throughout the history of literature but it used to be called by several names: speculative romance, pseudo-scientific romance (a term that sets a science fiction writer’s teeth on edge), utopian literature, fantasy — or, more frequently, given no name, simply lumped in with all other fiction.

But the term “science fiction” is now part of the language, as common as the neologism “guided missile.” We are stuck with it and I will use it … although personally I prefer the term “speculative fiction” as being more descriptive. I will use these two terms interchangeably, one being the common handle, the other being one that aids me in thinking — but with the same referent in each case.

“Science fiction” means different things to different people. “When I make a word do a lot of work like that,” said Humpty Dumpty, “I always pay it extra” — in which case the term science fiction has piled up a lot of expensive overtime. Damon Knight, a distinguished critic in this field, argues that there is no clear distinction between fantasy and science fiction, in which opinion August Derleth seems to agree. I cannot forcefully disagree with their lines of reasoning — but I wonder if they have made their definitions so broad as to include practically all fiction? To define is to limit: a definition cannot be useful unless it limits. Certainly Mickey Spillane’s murder stories could easily be classed as fantasies, as can many or most of the love stories appearing in the big slick magazines. But I feel sure that Mr. Knight and Mr. Derleth did not intend their definitions to be quite that unbounded and in any case my difference of opinion with them is merely a matter of taste and personal convenience.

Theodore Sturgeon, a giant in this field, defines a science fiction story as one in which the story would not exist if it were not for the scientific element — an admirably sharp delimitation but one which seems to me perhaps as uncomfortably tight as the one above seems to me unusefully roomy. It would exclude from the category “science fiction” much of Mr. Sturgeon’s best work, stories which are to my mind speculative rather than fantastic. There are many stories that are lumped into the class “science fiction” in the minds of most people (and in mine) which contain only a detectable trace, or none, of science — for example, Sinclair Lewis’ It Can’t Happen Here, Fritz Leiber’s great short story “Coming Attraction,” Thomas F. Tweed’s novel Gabriel Over the White House. All three stories are of manners and morals; any science in them is merely parsley trimming, not the meat. Yet each is major speculation, not fantasy, and each must be classed as science fiction as the term is commonly used.

Reginald Bretnor, author, editor and acute critic of this field, gives what is to me the most thoughtful, best reasoned, and most useful definition of science fiction. He sees it as a field of literature much broader than that most often termed “main-stream” literature — or “non-science fiction,” if you please — science fiction being that sort in which the author shows awareness of the nature and importance of the human activity known as the scientific method, shows equal awareness of the great body of human knowledge already collected through that activity, and takes into account in his stories the effects and possible future effects on human beings of scientific method and scientific fact. This indispensable three-fold awareness does not limit the science fiction author to stories about science — he need not write a gadget story; indeed a gadget story would not be science fiction under this definition if the author failed in this three-fold awareness. Any subject can be used in a science fiction story under this definition, provided (and indispensably required) that the author has the attitude comprised by the three-fold awareness and further provided that he has and uses appropriately that body of knowledge pertinent to the scope of his story. I have paraphrased in summary Mr. Bretnor’s comments and I hope he will forgive me.

Mr. Bretnor’s definition gives the science fiction author almost unlimited freedom in subject matter while requiring of him high, rigorous, and mature standards in execution.

In contrast to science fiction thus defined, non-science fiction — all other fiction including the most highly acclaimed “literary” novels — at most shows awareness of the by-products of scientific method already in existence. Non-science fiction admits the existence of the automobile, radar, polio vaccine, H-bombs, etc., but refuses to countenance starships and other such frivolities. That is to say, non-science fiction will concede that water is running down hill but refuses to admit that it might ever reach the bottom … or could ever be pumped up again. It is a static attitude, an assumption that what is now forever shall be.