This semester, I am the TA for Scientific Writing (Ch/ChE 91) at Caltech. As you can guess, this course is about... well... writing, specifically writing in the sciences. In the course, students will do what scientists do: write, review, and revise (both a scientific perspective and a research article).
But, before anyone in the sciences can write, they have to read. A lot.
As an undergraduate, I learned the hard way that I needed to relearn how to read. After reading several scientific journal articles front-to-back, word-by-word, I was bored out of my mind and found that I was unable to recall any of the information (I thought) I had just taken in.
As it turns out, there's more to reading a scientific (or any other academic) article than picking it up and reading the words on the page. The process of choosing an article alone can be tedious and finding the right one(s) for you might be the most time-consuming part of "reading" papers. The good news is that, if you strategize how you pick out an article, you can better read the paper (and even might not have to read the whole paper).
Hitting the library or an online journal database to find something in your field of interest might seem like the obvious first step in reading a scientific paper. However, before you do that, it is imperative that you ask yourself why you are embarking on a quest to sift through someone else's dense text and confusing captions as well as what information you are trying to find.
There are several reasons to consult the extensive databases of journal articles (just imagine if we didn't have the Internet), including coursework, research, and journalism. You might be seeking a research article because it is required reading for a class. You might need information for a grant proposal or for writing your own paper. You might be interested in turning a lump of dense jargon and undecipherable equations into something the general public can understand. Regardless of the purpose, it is important to identify what kind of information you need.
There are numerous ways to break down types of articles hence methods for gathering information. If you are writing your own research article, you need plenty of background information and previous results, so you probably want to focus on the results and discussions of papers. If you are writing a grant proposal, you might focus on methods upon which you can improve as well as introductions to figure out the current state of your field. If you are writing a scientific perspective, like Ch/ChE 91's first assignment, you probably need to focus on both introductory information (to figure out what the state of the field was before) and the results (to figure out the state after), but only for one or two papers.
Once you've established the types of information you need to read, you have to actually pick out a paper. Decide which journals you want to consult (if you are interested in astronomy you probably don't need to be looking at microbiology journals) and figure out some key terms. You don't want to look through the thousands of titles that are listed if you search for "polycyclic aromatic hydrocarbons" in Web of Science, and even if you did want to, you just don't have the time.
Look at the titles of the articles you are considering. Luckily for you, scientific articles typically have horrendously boring titles that might as well be an abstract of the paper's abstract. You can get a lot of information from a title and can quickly sort between the "maybe" articles and the "definitely nots."
After you have a smaller pile of papers, start with the abstracts. An abstract of a paper (should) give you a nice summary of the area of the work and any important points regarding the motivation, methods, and results of the project. If you cannot understand the abstract, consider tossing that paper aside; the actual text of the paper probably won't be any clearer. You can get a lot of information from the abstracts, and sometimes it contains the only information you'll actually need (but you shouldn't stop there... most of the time).
From the abstract, reading a paper is (sort of) a breeze. Start with the introduction. The intro to a paper will give you lots of useful background information for what is happening in a given area of research, and it also provides lots of citations of papers for you to consider next when you realize the first article you picked out makes a great sleeping aide. Look at the summary (sometimes called the conclusions section) at the very end. There, you'll find the big take-aways of the article and perhaps the outlook of the project (what's happening next and why you should care). Unless you need details for your own paper, you can probably just skip the methods section and skim the results and discussion sections.
While you read, take notes. Ask questions. Try to understand the figures. Ask questions. Finally, be flexible; you might read differently every time especially if you are reading for different reasons.
How you go about these things depends on your own learning and reading style, as well as the type of reading you are trying to accomplish. If you want some ideas for methods of how to read, check out "How to Read (Science) Journal Articles" and the update one year later on Caffeinated Confidence, "How to read and understand a scientific paper: A guide for non-scientists" by Jennifer Raff (Violent Metaphors), and "How to Read a Scientific Paper" from Science Buddies. Also check out these guides from Rice and Stanford.