使用. AsParallel()。AsOrdered()而不是Parallel.ForEach()。这允许并行处理项，并且仍然按照与它们在输入序列中出现的顺序相同的顺序枚举处理的项。

var inputItems = new List<int>(new[] { 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 });//List<int> is just example, input sequence can be any IEnumerable<T>

var processedItems = inputItems
    .AsParallel()//Allow parallel processing of items
    .AsOrdered()//Force items in output enumeration to be in the same order as in input
    .WithMergeOptions(ParallelMergeOptions.NotBuffered)//Allows enumeration of processed items as soon as possible (before all items are processed) at the cost of slightly lower performace
    .Select(item =>
        {
            //Do some processing of item
            Console.WriteLine("Processing item " + item);

            return item;//return either input item itself, or processed item (e.g. item.ToString())
        });

//You can use processed enumeration just like any other enumeration (send it to the customer, enumerate it yourself using foreach, etc.), items will be in the same order as in input enumeration.
foreach (var processedItem in processedItems)
{
    //Do whatever you want with processed item
    Console.WriteLine("Enumerating item " + processedItem);
}

如果你非常关心顺序，你可以使用像ConcurrentQueue<T>这样的东西，但是这是FIFO逻辑，所以你需要在删除项目时小心。
如果您正在寻找一个完全灵活的选项，还有BlockingCollection<T>，在这种情况下非常方便。Here是一篇很好的文章，this是原始的MS文档。
下面你可以看到一个尝试从列表中获取一个项目的案例：

BlockingCollection<MyClass> bCollection = new BlockingCollection<MyClass>(boundedCapacity: 2);
bCollection.Add(new MyClass{ Field1 = "Test" });
bCollection.Add(new MyClass{ Field1 = "Test2" };

var item = bCollection.Take();
item = bCollection.Take();

if (bCollection.TryTake(out item, TimeSpan.FromSeconds(1)))
{
    Console.WriteLine(item);
}
else
{
    Console.WriteLine("No item removed");
}