请看下图(大图)中每个类在类层次结构中的位置。TreeMap实现了SortedMap和NavigableMap，而HashMap没有。

HashTable已经过时，应该使用相应的ConcurrentHashMap类。

虽然这里有很多很好的答案，但我想给出我自己的表，描述与Java 11绑定的各种Map实现。

我们可以在图表中看到这些差异:

HashMap is the general-purpose Map commonly used when you have no special needs. LinkedHashMap extends HashMap, adding this behavior: Maintains an order, the order in which the entries were originally added. Altering the value for key-value entry does not alter its place in the order. TreeMap too maintains an order, but uses either (a) the “natural” order, meaning the value of the compareTo method on the key objects defined on the Comparable interface, or (b) invokes a Comparator implementation you provide. TreeMap implements both the SortedMap interface, and its successor, the NavigableMap interface. NULLs: TreeMap does not allow a NULL as the key, while HashMap & LinkedHashMap do. All three allow NULL as the value. HashTable is legacy, from Java 1. Supplanted by the ConcurrentHashMap class. Quoting the Javadoc: ConcurrentHashMap obeys the same functional specification as Hashtable, and includes versions of methods corresponding to each method of Hashtable.

@Amit: SortedMap is an interface whereas TreeMap is a class which implements the SortedMap interface. That means if follows the protocol which SortedMap asks its implementers to do. A tree unless implemented as search tree, can't give you ordered data because tree can be any kind of tree. So to make TreeMap work like Sorted order, it implements SortedMap ( e.g, Binary Search Tree - BST, balanced BST like AVL and R-B Tree , even Ternary Search Tree - mostly used for iterative searches in ordered way ).

public class TreeMap<K,V>
extends AbstractMap<K,V>
implements SortedMap<K,V>, Cloneable, Serializable

在坚果壳 HashMap:给出O(1)的数据，没有排序

TreeMap:给出O(log N)，以2为底的数据。使用有序键

LinkedHashMap:是具有链表(想想索引- skiplist)功能的哈希表，以插入树的方式存储数据。最适合实现LRU(最近最少使用)。

这三个都表示从唯一键到值的映射，因此实现了Map接口。

HashMap is a map based on hashing of the keys. It supports O(1) get/put operations. Keys must have consistent implementations of hashCode() and equals() for this to work. LinkedHashMap is very similar to HashMap, but it adds awareness to the order at which items are added (or accessed), so the iteration order is the same as insertion order (or access order, depending on construction parameters). TreeMap is a tree based mapping. Its put/get operations take O(log n) time. It requires items to have some comparison mechanism, either with Comparable or Comparator. The iteration order is determined by this mechanism.

它们都提供一个key->值映射和一种遍历键的方法。最重要的区别 这些类是时间保证和键的顺序。

HashMap offers 0(1) lookup and insertion. If you iterate through the keys, though, the ordering of the keys is essentially arbitrary. It is implemented by an array of linked lists. TreeMap offers O(log N) lookup and insertion. Keys are ordered, so if you need to iterate through the keys in sorted order, you can. This means that keys must implement the Comparable interface.TreeMap is implemented by a Red-Black Tree. LinkedHashMap offers 0(1) lookup and insertion. Keys are ordered by their insertion order. It is implemented by doubly-linked buckets.

假设你将一个空的TreeMap, HashMap和LinkedHashMap传递给下面的函数:

void insertAndPrint(AbstractMap<Integer, String> map) {
  int[] array= {1, -1, 0};
  for (int x : array) {
    map.put(x, Integer.toString(x));
  }
  for (int k: map.keySet()) {
   System.out.print(k + ", ");
  }
}

它们的输出如下所示。

对于HashMap，在我自己的测试中，输出是{0,1，-1}，但它可以是任何顺序。没有任何保证 排序。 Treemap，输出为{- 1,0,1} LinkedList，输出为{1，-1,0}

HashMap: 不维持秩序 比LinkedHashMap快 用于存储对象堆 LinkedHashMap: LinkedHashMap插入顺序将被维持 比HashMap慢，比TreeMap快 如果你想保持一个插入顺序，使用这个。 TreeMap: TreeMap是一种基于树的映射 TreeMap将遵循键的自然顺序 比HashMap和LinkedHashMap慢 在需要维护自然(默认)排序时使用TreeMap