How can I calculate the difference between two ArrayLists?

In Java, you can use the Collection interface's removeAll method.

// Create a couple ArrayList objects and populate them
// with some delicious fruits.
Collection firstList = new ArrayList() {{
    add("apple");
    add("orange");
}};

Collection secondList = new ArrayList() {{
    add("apple");
    add("orange");
    add("banana");
    add("strawberry");
}};

// Show the "before" lists
System.out.println("First List: " + firstList);
System.out.println("Second List: " + secondList);

// Remove all elements in firstList from secondList
secondList.removeAll(firstList);

// Show the "after" list
System.out.println("Result: " + secondList);

The above code will produce the following output:

First List: [apple, orange]
Second List: [apple, orange, banana, strawberry]
Result: [banana, strawberry]

You already have the right answer. And if you want to make more complicated and interesting operations between Lists (collections) use apache commons collections (CollectionUtils) It allows you to make conjuction/disjunction, find intersection, check if one collection is a subset of another and other nice things.

In Java 8 with streams, it's pretty simple actually. EDIT: Can be efficient without streams, see lower.

List<String> listA = Arrays.asList("2009-05-18","2009-05-19","2009-05-21");
List<String> listB = Arrays.asList("2009-05-18","2009-05-18","2009-05-19","2009-05-19",
                                   "2009-05-20","2009-05-21","2009-05-21","2009-05-22");

List<String> result = listB.stream()
                           .filter(not(new HashSet<>(listA)::contains))
                           .collect(Collectors.toList());

Note that the hash set is only created once: The method reference is tied to its contains method. Doing the same with lambda would require having the set in a variable. Making a variable is not a bad idea, especially if you find it unsightly or harder to understand.

You can't easily negate the predicate without something like this utility method (or explicit cast), as you can't call the negate method reference directly (type inference is needed first).

private static <T> Predicate<T> not(Predicate<T> predicate) {
    return predicate.negate();
}

If streams had a filterOut method or something, it would look nicer.

Also, @Holger gave me an idea. ArrayList has its removeAll method optimized for multiple removals, it only rearranges its elements once. However, it uses the contains method provided by given collection, so we need to optimize that part if listA is anything but tiny.

With listA and listB declared previously, this solution doesn't need Java 8 and it's very efficient.

List<String> result = new ArrayList(listB);
result.removeAll(new HashSet<>(listA));

EDIT: Original question did not specify language. My answer is in C#.

You should instead use HashSet for this purpose. If you must use ArrayList, you could use the following extension methods:

var a = arrayListA.Cast<DateTime>();
var b = arrayListB.Cast<DateTime>();    
var c = b.Except(a);

var arrayListC = new ArrayList(c.ToArray());

using HashSet...

var a = new HashSet<DateTime>(); // ...and fill it
var b = new HashSet<DateTime>(); // ...and fill it
b.ExceptWith(a); // removes from b items that are in a

Although this is a very old question in Java 8 you could do something like

 List<String> a1 = Arrays.asList("2009-05-18", "2009-05-19", "2009-05-21");
 List<String> a2 = Arrays.asList("2009-05-18", "2009-05-18", "2009-05-19", "2009-05-19", "2009-05-20", "2009-05-21","2009-05-21", "2009-05-22");

 List<String> result = a2.stream().filter(elem -> !a1.contains(elem)).collect(Collectors.toList());

I guess you're talking about C#. If so, you can try this

    ArrayList CompareArrayList(ArrayList a, ArrayList b)
    {
        ArrayList output = new ArrayList();
        for (int i = 0; i < a.Count; i++)
        {
            string str = (string)a[i];
            if (!b.Contains(str))
            {
                if(!output.Contains(str)) // check for dupes
                    output.Add(str);
            }
        }
        return output;
    }

You are just comparing strings.

Put the values in ArrayList A as keys in HashTable A.
Put the values in ArrayList B as keys in HashTable B.

Then, for each key in HashTable A, remove it from HashTable B if it exists.

What you are left with in HashTable B are the strings (keys) that were not values in ArrayList A.

C# (3.0) example added in response to request for code:

List<string> listA = new List<string>{"2009-05-18","2009-05-19","2009-05-21'"};
List<string> listB = new List<string>{"2009-05-18","2009-05-18","2009-05-19","2009-05-19","2009-05-20","2009-05-21","2009-05-21","2009-05-22"};

HashSet<string> hashA = new HashSet<string>();
HashSet<string> hashB = new HashSet<string>();

foreach (string dateStrA in listA) hashA.Add(dateStrA);
foreach (string dateStrB in listB) hashB.Add(dateStrB);

foreach (string dateStrA in hashA)
{
    if (hashB.Contains(dateStrA)) hashB.Remove(dateStrA);
}

List<string> result = hashB.ToList<string>();