Lemma/Proposition/Theorem, which one should we pick? This is something that confuses me.
I have read a few mathematical texts and they often seem to use Lemma/Proposition/Theorem if they have a particular statement.
Now, which one to use? A lemma can be something you need to prove a more important theorem, but then what about Fatou's Lemma?
When to pick Proposition or Theorem?
 A: There seem to be two issues here.  One is why certain well-known results are called Lemmas, such as Zorn's, Yoneda's, Nakayama's, and so on.  I don't know the answer to this; presumably it is a mixture of what was written in some original source and the results of the transmission of that original source through the mathematical tradition.  (As one interesting example of how labels can be changed in the course of transmission, there is a result in the theory of automorphic forms and Galois representations, very well known to experts, universally referred to as "Ribet's Lemma"; however, in the original paper it is labelled as a proposition!)
The second issue is how contemporary writers label the results in their papers.  My experience is that typically the major results of the paper are called theorems, the lesser results are called propositions (these are typically ingredients in the proofs of the theorems which are also stand-alone statements that may be of independent interest), and the small technical results are called lemmas.   This probably varies quite a bit from writer to writer (and perhaps also from field to field?).  
A: I don't know if there are any hard and fast rules, but here is a rough start for others to nitpick:


*

*A Theorem is a major result that you care about (e.g. "the goal of this paper is to prove the following theorem").

*A Lemma is a useful result that needs to be invoked repeatedly to prove some Theorem or other. Note that sometimes Lemmas can become much more useful than the Theorems they were originally written down to prove. 

*A Proposition is a technical result that does not need to be invoked as often as a Lemma.

A: I'm by no means knowledgeable on this topic, but this article seems to have useful definitions.
What is the difference between a theorem, a lemma, and a corollary?
Quote:

Posted by DAVE RICHESON.
I prepared the following handout for my Discrete Mathematics class
(here’s a pdf version).
Definition — a precise and unambiguous description of the meaning of a
mathematical term.  It characterizes the meaning of a word by giving
all the properties and only those properties that must be true.
Theorem — a mathematical statement that is proved using rigorous
mathematical reasoning.  In a mathematical paper, the term theorem is
often reserved for the most important results.
Lemma — a minor result whose sole purpose is to help in proving a
theorem.  It is a stepping stone on the path to proving a theorem.
Very occasionally lemmas can take on a life of their own (Zorn’s
lemma, Urysohn’s lemma, Burnside’s lemma, Sperner’s lemma).
Corollary — a result in which the (usually short) proof relies heavily
on a given theorem (we often say that “this is a corollary of Theorem
A”).
Proposition — a proved and often interesting result, but generally
less important than a theorem.
Conjecture — a statement that is unproved, but is believed to be true
(Collatz conjecture, Goldbach conjecture, twin prime conjecture).
Claim — an assertion that is then proved.  It is often used like an
informal lemma.
Axiom/Postulate — a statement that is assumed to be true without
proof. These are the basic building blocks from which all theorems are
proved (Euclid’s five postulates, Zermelo-Fraenkel axioms, Peano
axioms).
Identity — a mathematical expression giving the equality of two (often
variable) quantities (trigonometric identities, Euler’s identity).
Paradox — a statement that can be shown, using a given set of axioms
and definitions, to be both true and false. Paradoxes are often used
to show the inconsistencies in a flawed theory (Russell’s paradox).
The term paradox is often used informally to describe a surprising or
counterintuitive result that follows from a given set of rules
(Banach-Tarski paradox, Alabama paradox, Gabriel’s horn).

A: From my reading in maths, I have found the distinction useful in organization and presentation of results. I find it hard to believe that Hardy would have favored a "theorem only" approach. The distinction I find useful is essentially the following:
Theorem - main result of the paper. One to three theorems per paper, unless a long paper with many sections, where one to three theorems per section may be appropriate.
Proposition - result that is used in the proofs, but which may or may not be proved in the current presentation, and for which no originality is claimed.
Lemma - technical result used in the proof of the theorem, which is claimed as original and proved, but the main interest in which lies its use in the proof of one or more theorems.
Corollary - a specialization of a just presented theorem, in terms more likely to be useful in practice, or of intuitive interest.
For example Zorn's Lemma on partially ordered chains having maximal elements is not of much interest in itself, but it is key to have that established before proving Hahn-Banach Theorem or Tychonoff's Theorem.
I believe that this categorization is actually very useful for the following reasons:
(1) It helps the reader understand the purpose of a result in the larger scheme of the presentation, and to differentiate between results that are to be identified with the paper/section/chapter as part of its raison d'etre, versus ancillary results that may be important but are only being formally identified for use, proved or not, and not claimed.
(2) If one sticks to theorems only, particularly if one is not numbering within section, then one quickly reached double digit theorems, and there is potential cognitive interference between the section numbers and the decimal notation.
I'm surprised this type of thing doesn't have a well known codification somewhere ... it probably does, but I thought I would offer these arguments in favor of the distinction since the thread consensus seems to be trending in the other direction.
A: While generally  the terms are used as suggested by Qiaochu, there are some authors who are bothered by these nebulous subjective terms. For example, Kaplansky wrote in the preface of his classic textbook Commutative Rings

In the style of Landau, or Hardy and Wright, I have presented the material as an unbroken series of theorems. I prefer this to the n-place decimal system favored by some authors, and I have also grown tired of seeing a barrage of lemmas, propositions, corollaries, and scholia (whatever they are). I admit that this way the lowliest lemma gets elevated to the same eminence as the most awesome theorem. Also, the number of theorems becomes impressive,  so impressive that I felt the need to add an index of theorems. 

A: There are lots of philosophical answers for this question. but the best answer is the easiest one that can always be remembered for mathematics courses:

If we can prove a statement true, then that statement is called a
proposition. A proposition of major importance is called a theorem.
A theorem cannot be proved by example; however, the standard way to
show that a statement is not a theorem is to provide a counterexample.
Theorems are tools that make new and productive applications of
mathematics possible. We use examples to give insight into existing
theorems and to foster intuitions as to what new theorems might be
true. Applications, examples, and proofs are tightly interconnected
much more so than they may seem at first appearance.
Sometimes instead of proving a theorem or proposition all at once, we
break the proof down into modules; that is, we prove several
supporting propositions, which are called lemmas, and use the
results of these propositions to prove the main result. If we can
prove a proposition or a theorem, we will often, with very little
effort, be able to derive other related propositions called
corollaries.

From Thomas Judson.
A: Ultimately, these are tools to make a paper more readable. I have found it useful to use the terms proposition and lemma to separate different threads in a paper. Lemmas are very much in line with the theorem (helping to display progression towards the theorem). I don't agree with Matt E's answer that a lemma should be small and technical, it may be deep and interesting in its own right, but to make it a theorem would distract from the main narrative of the paper. I agree that propositions may be less aligned, more  ad hoc, and hence of independent interest. If propositions are not explicitly used later in the paper, but serve to highlight something important, then perhaps they should be called remarks. 
A: I believe the answers and comments to this question already show the conclusion:

*

*The Theorem should deliver an important results, or the main results in a paper.


*The use of Lemma/proposition are subjective to the writers. The answers and comments (in, e.g., @Qiaochu Yuan) show that there is no common agreement on what is the correct use.


*A corollary is apparently a derivative results of another result.
