这是一个“圣战”式的问题。换句话说，两种选择都是好的，但人们会有自己的偏好，他们会誓死捍卫这些偏好。

给杰森·科恩的论点投了一票，因为它表现得很好。让我一步一步地分解它。: -)

The NPE JavaDoc explicitly says, "other illegal uses of the null object". If it was just limited to situations where the runtime encounters a null when it shouldn't, all such cases could be defined far more succinctly. Can't help it if you assume the wrong thing, but assuming encapsulation is applied properly, you really shouldn't care or notice whether a null was dereferenced inappropriately vs. whether a method detected an inappropriate null and fired an exception off. I'd choose NPE over IAE for multiple reasons It is more specific about the nature of the illegal operation Logic that mistakenly allows nulls tends to be very different from logic that mistakenly allows illegal values. For example, if I'm validating data entered by a user, if I get value that is unacceptable, the source of that error is with the end user of the application. If I get a null, that's programmer error. Invalid values can cause things like stack overflows, out of memory errors, parsing exceptions, etc. Indeed, most errors generally present, at some point, as an invalid value in some method call. For this reason I see IAE as actually the MOST GENERAL of all exceptions under RuntimeException. Actually, other invalid arguments can result in all kinds of other exceptions. UnknownHostException, FileNotFoundException, a variety of syntax error exceptions, IndexOutOfBoundsException, authentication failures, etc., etc.

总的来说，我觉得NPE受到了很大的诋毁，因为传统上一直与未能遵循快速失效原则的代码联系在一起。再加上JDK未能用消息字符串填充NPE，这确实产生了一种强烈的负面情绪，这种情绪是没有根据的。实际上，从运行时的角度来看，NPE和IAE之间的区别仅限于名称。从这个角度来看，你的名字越精确，你给调用者的信息就越清晰。

如果你不希望null是一个允许的值，似乎会调用IllegalArgumentException，如果你试图使用一个结果为null的变量，则会抛出NullPointerException。

抛出一个排除空参数的异常(无论是NullPointerException还是自定义类型)使得自动化空测试更加可靠。这种自动化测试可以通过反射和一组默认值来完成，就像在Guava的NullPointerTester中一样。例如，NullPointerTester将尝试调用以下方法…

Foo(String string, List<?> list) {
  checkArgument(string.length() > 0);
  // missing null check for list!
  this.string = string;
  this.list = list;
}

...with two lists of arguments: "", null and null, ImmutableList.of(). It would test that each of these calls throws the expected NullPointerException. For this implementation, passing a null list does not produce NullPointerException. It does, however, happen to produce an IllegalArgumentException because NullPointerTester happens to use a default string of "". If NullPointerTester expects only NullPointerException for null values, it catches the bug. If it expects IllegalArgumentException, it misses it.

Apache Commons Lang有一个NullArgumentException，它完成了这里讨论的许多事情:它扩展了IllegalArgumentException，并且它唯一的构造函数采用了参数的名称，而参数的名称本应该是非空的。

虽然我觉得抛出NullArgumentException或IllegalArgumentException之类的异常更准确地描述了异常情况，但我和同事们还是选择遵从Bloch在这个问题上的建议。

当试图访问具有当前值为null的引用变量的对象时，抛出NullPointerException。

当方法接收到格式与方法预期不同的参数时，抛出IllegalArgumentException。