Very important.
A lot of the mental status exam should be based on the patient's body language. In forensic psychiatry this is especially important because many patients malinger. You cannot often take what the evaluee says at face value.
Body language, like most aspects in the mental health sciences should be based on 2 models: behavior based on the general population (and how this applies to the person), and the person's own idiosyncratic behavior that may not be seen in others in the population.
In violent people, there's one model that analyzes the person's behavior. This model emphasizes categorizing the person as perpetrator of affective violence vs. predatory violence. In this model, body language is very important. In affective violence, the person is noticeably angered (angry affect, increased BP, possible sweating, threats, defensive/offensive body posture). In predatory violence, the person is not noticeably angered, and is seen as calm before and after the attack.
In perpetrators of predatory violence, determining safety is especially difficult phenomenon because it's much more difficult to tell if the person will attack another ahead of the attack.
Its difficult to look for lies when you are telling them yourself. No offence intended but just look at the lies we might be telling and sometimes with good reason.
According to studies, psychiatrists were not able to demonstrate we were any better at telling if someone was lying than a college student. The college student is by no means a complement to us (as if they're better educated than others) because college students were the control population in the study. The only group to show a statistical improvement over the college students were law enforcement that regularly interrogated suspects. (I need to double check the exact profession because it's been a few months since I read the study).
For that reason, we in the mental health profession should not be in a position to tell if someone is telling the truth or not unless we can point to specific obejctive reasons (E.g. psychological testing, contradicting statements made by the person, atypical signs and symptoms).
Do not ever think that you can tell if someone is lying simply because you are a psychiatrist. I would believe that a psychiatrist, who knows a person very very well could tell better than a random person, but that's not because the psychiatrist is somehow better than anyone else. It's because anyone who knows someone else very very well would be able to tell better.