This is a bug, manifest in all versions since 5.2.0. Actually it is a "multibug", exhibiting a series of surprising inconsistencies. * That null is silently ignored isn't documented (not for operators '<' and '>'). * One might expect 0 to be used as the only reasonable default, but it isn't. * One might any expression such as a < count < b involving a null Integer 'count' to return false, but this isn't so either. One might think that the default is "any integer in [a+1,b-1]" - but read on! * Surprisingly, an expression with (count == null) such as 1 < count < 1 or 10 < count < 1 will always fail, but 1 < count < 2 succeeds, although there is no integer that would ever make this expression true. * One would really (REALLY!) expect that 1 < count < 100 and 1 < count && count < 100 would always produce the same result, no matter what the value of count is, but for null this isn't so. * One would really expect that count <= Integer.MAX_VALUE would always return true, given that it doesn't throw a NPE, but it returns true for count == null. But, sure enough, Integer.MIN_VALUE <= count <= Integer.MAX_VALUE always returns true. In other words, if you have an Integer that might be null, always add a test against null. -W On 26/09/2012, mpgong <michael.p.gong(a)lmco.com&gt; wrote: