Check for null in foreach loop

Assuming that the type of elements in file.Headers is T you could do this

foreach(var header in file.Headers ?? Enumerable.Empty<T>()){
  //do stuff
}

this will create an empty enumerable of T if file.Headers is null. If the type of file is a type you own I would, however, consider changing the getter of Headers instead. null is the value of unknown so if possible instead of using null as "I know there are no elements" when null actually(/originally) should be interpreted as "I don't know if there are any elements" use an empty set to show that you know there are no elements in the set. That would also be DRY'er since you won't have to do the null check as often.

EDIT as a follow up on Jons suggestion, you could also create an extension method changing the above code to

foreach(var header in file.Headers.OrEmptyIfNull()){
  //do stuff
}

In the case where you can't change the getter, this would be my own preferred since it expresses the intention more clearly by giving the operation a name (OrEmptyIfNull)

The extension method mentioned above might make certain optimizations impossible for the optimizer to detect. Specifically, those that are related to IList using method overloading this can be eliminated

public static IList<T> OrEmptyIfNull<T>(this IList<T> source)
{
    return source ?? Array.Empty<T>();
}

Frankly, I advise: just suck up the null test. A null test is just a brfalse or brfalse.s; everything else is going to involve much more work (tests, assignments, extra method calls, unnecessary GetEnumerator(), MoveNext(), Dispose() on the iterator, etc).

An if test is simple, obvious, and efficient.

Just as a slight cosmetic addition to Rune's suggestion, you could create your own extension method:

public static IEnumerable<T> OrEmptyIfNull<T>(this IEnumerable<T> source)
{
    return source ?? Enumerable.Empty<T>();
}

Then you can write:

foreach (var header in file.Headers.OrEmptyIfNull())
{
}

Change the name according to taste :)