Primary source materials should be evaluated as you would any other source.
Things to consider when looking at a primary source include:
Each person brings a certain amount of bias to what they create, even when trying to create something neutral. This bias doesn't make the material useless to a researcher, but should be considered when interpreting and using the source.
For example, is the material a piece of propaganda designed to promote a specific cause?
Was the item written by someone in the military versus a civilian?
A diary written by someone on the American side of the Revolutionary war versus by someone on the British side of the war.
A speech by someone who is an abolitionist versus a slave owner.
A newspaper that is written in the United States versus one written in France.
A Nazi's description of the Holocaust versus that of a survivor of a concentration camp.
A treaty created by the winner of a war versus a treaty created by two countries on equal footing.
If you can see/read the original item in the original language that would be the best thing; however, it is not always possible.
If it is a copy, reprint, or translated version of the original there could be mistakes made in this version.
Try checking more than one translation - are there differences?
Did the author write this for him/herself, such as personal diary?
Was it written for a specific person to read, such as a letter? If so, what was the relationship between the two people?
Was it a speech designed to sway people to vote a certain way?
An autobiography or memoir, while still a useful source, would have been written after the fact, not during it.
Things like the passage of time, memory changes, and events that occurred after the event could affect the work.
A letter written in the trenches of war would likely be very different than one written after the battle was over.
A news report by an embedded reporter filming during a firefight versus a recounting of the fight by the reporter after the shooting has ended will have a very different tone.
Can you believe everything someone writes or says? Of course not.
It can help to know a little about the person who wrote the item. Are they considered to be an exaggerator (like Captain John Smith or Mark Twain) or more of an honest person, unlikely to tell lies, (George Washington or Abraham Lincoln.)
Know some basic facts about an event - if someone claims to have been a soldier in World War II but wasn't born until 1943, would you still believe their story?
As is often evident in eye witness accounts one person can be sure the car was red and another that it was green. Consider that a statement made after a traumatic event may be inaccurate, not necessarily on purpose but due to the person's state of mind at the time.
Consider material found online carefully.
Look for the about page to see who is behind the page or organization.
Try using reputable sources such as the Library of Congress, the British Library, the United Nations or an archival collection provided by a University or government organization.
See also: Evaluating Web Information