Commenter Liz asked yesterday about personal statements for grad school and it struck me as something I could write a quick response to.
First the don'ts
As I mentioned yesterday, don't write about your child science inspiration. A letter that starts out "when I was young..." immediately makes me want to stab my eyes out.
As Odyssey pointed out, don't use personal experience as a motivation to stamp out disease. We get it, lot's of people have diseases.
Don't relate your passion for science by including quotes from Nietzsche or Mother Goose. Seen both. True story.
Don't think that length = impressive. Usually it = boring as hell and only partially read.
Don't describe a detailed research plan for what you think you want to do.
As for the Do's
Realize who your audience is. What does a PI want to see in your letter? Competence as a writer, demonstrated experience in research, ability to concisely describe one's research (h/t bugdoc) and evidence that you have done some homework when it comes to what you would like to do in grad school.
That's pretty much it. It doesn't have to be a work of literary genius, but it should be clear and concise. Get across that you have research experience, you have learned something about how science works and that you want in on the deal.
Finally, look like you considered what a particular program or lab can offer. If you are applying to a rotation, write about the strength of the department in the field you are interested. Perhaps you enjoyed the 2010 paper by Dr. Schnapps on Ethanol treatment of Care Bears and would be interested in discussing future directions with this work. And maybe you were impressed by Dr. Gliter's 2011 work on unicorn rainbow jumping and would appreciate the opportunity to discuss projects applying unicorn jumping behavior to other meteorological conditions. If you are applying directly to a lab, then talk about the work being done in that lab and where you would be interested in contributing. Again, remember that proposing your own line of research that fits with where you think the lab is going is not a great strategy.
Above all, be organized, prepared and concise. And don't talk about your childhood.