Some people (apparently with a better sense of smell than I have) swear that it's important to eat what the enemy eats, wash with what the enemy washes and not to smoke if you shall go on a patrol in a forest. Your different smell could otherwise give away your presence to the enemy.
That may very well be true. The sense of smell has nevertheless a very restricted utility for armies.
Again and again and again some developers attempt to develop some explosive sniffer tool or similar for the military. Even worse; they even get funds for this nonsense. Here's a recent example: Danger Room: "Secret New Sensors Sniff For Afghanistan's fertilizer Bombs"
Let's be frank (even more than usual): That's a waste, crap, a rip-off. Such sensors are useless. To grant funds for such projects is an evidence for stupidity and incompetence.
A simple, short anecdote from WW2 can explain (and prove) it:
Soviet partisans mined railways to destroy German supply trains. The Germans reacted by deploying dog teams with dogs trained to sniff for the mines. They found some mines. The partisans observed this and dispersed small quantities of explosives almost everywhere on the tracks. The great sense of smell of the dogs was rendered useless, utterly useless. The false alarm rate made the whole "sniffing" approach 100% useless. It's still 100% useless.
Engineers who attempt to develop smell-based explosive detection tools for general or route reconnaissance are wasting time and money. This is especially obvious if it's not done in perfect secrecy.