Arrays make sense where the exact number of items will be known, and where searching by index makes sense. For example, if I wanted to store the exact state of my video output at a given moment without compression I would probably use an array of size [1024][768]. This will provide me with exactly what I need, and a list would be much, much slower to get the value of a given pixel. In places where an array does not make sense there are generally better data types than a list to deal with data effectively.

维基百科上有很好的章节介绍了它们的区别。

Linked lists have several advantages over arrays. Elements can be inserted into linked lists indefinitely, while an array will eventually either fill up or need to be resized, an expensive operation that may not even be possible if memory is fragmented. Similarly, an array from which many elements are removed may become wastefully empty or need to be made smaller. On the other hand, arrays allow random access, while linked lists allow only sequential access to elements. Singly-linked lists, in fact, can only be traversed in one direction. This makes linked lists unsuitable for applications where it's useful to look up an element by its index quickly, such as heapsort. Sequential access on arrays is also faster than on linked lists on many machines due to locality of reference and data caches. Linked lists receive almost no benefit from the cache. Another disadvantage of linked lists is the extra storage needed for references, which often makes them impractical for lists of small data items such as characters or boolean values. It can also be slow, and with a naïve allocator, wasteful, to allocate memory separately for each new element, a problem generally solved using memory pools.

http://en.wikipedia.org/wiki/Linked_list

A widely unappreciated argument for ArrayList and against LinkedList is that LinkedLists are uncomfortable while debugging. The time spent by maintenance developers to understand the program, e.g. to find bugs, increases and IMHO does sometimes not justify the nanoseconds in performance improvements or bytes in memory consumption in enterprise applicatons. Sometimes (well, of course it depends on the type of applications), it's better to waste a few bytes but have an application which is more maintainable or easier to understand.

例如，在Java环境中，使用Eclipse调试器，调试ArrayList将显示一个非常容易理解的结构:

arrayList   ArrayList<String>
  elementData   Object[]
    [0] Object  "Foo"
    [1] Object  "Foo"
    [2] Object  "Foo"
    [3] Object  "Foo"
    [4] Object  "Foo"
    ...

另一方面，查看LinkedList的内容并找到特定的对象变成了一个展开树的噩梦，更不用说过滤LinkedList内部信息所需的认知开销:

linkedList  LinkedList<String>
    header  LinkedList$Entry<E>
        element E
        next    LinkedList$Entry<E>
            element E   "Foo"
            next    LinkedList$Entry<E>
                element E   "Foo"
                next    LinkedList$Entry<E>
                    element E   "Foo"
                    next    LinkedList$Entry<E>
                    previous    LinkedList$Entry<E>
                    ...
                previous    LinkedList$Entry<E>
            previous    LinkedList$Entry<E>
        previous    LinkedList$Entry<E>

这实际上是一个效率问题，在链表中插入、删除或移动(而不是简单地交换)元素的开销是最小的，即操作本身是O(1)，而不是O(n)。如果您大量操作数据列表，这可能会产生显著的差异。您可以根据对数据类型的操作方式选择数据类型，并为所使用的算法选择最有效的数据类型。

首先，在c++中，使用链表应该不会比使用数组更麻烦。对于链表，可以使用std::list或boost指针列表。链表与数组的关键问题是指针需要额外的空间和糟糕的随机访问。你应该使用链表，如果你

你不需要随机访问数据 您将添加/删除元素，特别是列表中间的元素

数组Vs链表:

由于内存碎片，阵列内存分配有时会失败。 在数组中缓存更好，因为所有元素都分配了连续的内存空间。 编码比数组更复杂。 与数组不同，在链表上没有大小限制 在链表中插入/删除更快，在数组中访问更快。 从多线程的角度来看，链表更好。