Criteria是一个面向对象的API，而HQL意味着字符串连接。这意味着面向对象的所有好处都适用:

All else being equal, the OO version is somewhat less prone to error. Any old string could get appended into the HQL query, whereas only valid Criteria objects can make it into a Criteria tree. Effectively, the Criteria classes are more constrained. With auto-complete, the OO is more discoverable (and thus easier to use, for me at least). You don't necessarily need to remember which parts of the query go where; the IDE can help you You also don't need to remember the particulars of the syntax (like which symbols go where). All you need to know is how to call methods and create objects.

由于HQL非常像SQL(大多数开发人员已经非常了解SQL)，因此这些“不必记住”参数就没有那么重要了。如果HQL有更大的不同，那么这就更重要了。

为了两全其美，HQL的表达性和简洁性以及Criteria的动态特性可以考虑使用Querydsl。

Querydsl支持JPA/Hibernate, JDO, SQL和Collections。

我是Querydsl的维护者，所以这个答案是有偏见的。

We used mainly Criteria in our application in the beginning but after it was replaced with HQL due to the performance issues. Mainly we are using very complex queries with several joins which leads to multiple queries in Criteria but is very optimized in HQL. The case is that we use just several propeties on specific object and not complete objects. With Criteria the problem was also string concatenation. Let say if you need to display name and surname of the user in HQL it is quite easy (name || ' ' || surname) but in Crteria this is not possible. To overcome this we used ResultTransormers, where there were methods where such concatenation was implemented for needed result. Today we mainly use HQL like this:

String hql = "select " +
            "c.uuid as uuid," +
            "c.name as name," +
            "c.objective as objective," +
            "c.startDate as startDate," +
            "c.endDate as endDate," +
            "c.description as description," +
            "s.status as status," +
            "t.type as type " +
            "from " + Campaign.class.getName() + " c " +
            "left join c.type t " +
            "left join c.status s";

Query query =  hibernateTemplate.getSessionFactory().getCurrentSession().getSession(EntityMode.MAP).createQuery(hql);
query.setResultTransformer(Transformers.ALIAS_TO_ENTITY_MAP);
return query.list();

因此，在本例中，返回的记录是所需属性的映射。

HQL和criteriaQuery在性能上是有区别的，每次你使用criteriaQuery发起查询时，它会为表名创建一个新的别名，这个别名不会反映在任何DB的最后一个查询缓存中。这会导致编译生成的SQL的开销，需要更多的时间来执行。

关于抓取策略[http://www.hibernate.org/315.html]

Criteria respects the laziness settings in your mappings and guarantees that what you want loaded is loaded. This means one Criteria query might result in several SQL immediate SELECT statements to fetch the subgraph with all non-lazy mapped associations and collections. If you want to change the "how" and even the "what", use setFetchMode() to enable or disable outer join fetching for a particular collection or association. Criteria queries also completely respect the fetching strategy (join vs select vs subselect). HQL respects the laziness settings in your mappings and guarantees that what you want loaded is loaded. This means one HQL query might result in several SQL immediate SELECT statements to fetch the subgraph with all non-lazy mapped associations and collections. If you want to change the "how" and even the "what", use LEFT JOIN FETCH to enable outer-join fetching for a particular collection or nullable many-to-one or one-to-one association, or JOIN FETCH to enable inner join fetching for a non-nullable many-to-one or one-to-one association. HQL queries do not respect any fetch="join" defined in the mapping document.

对于动态查询，我更喜欢使用条件查询。例如，根据某些参数，动态地添加一些排序或保留一些部分(例如限制)要容易得多。

另一方面，我使用HQL进行静态和复杂的查询，因为它更容易理解/阅读HQL。此外，我认为HQL更强大一些，例如对于不同的连接类型。

对于动态条件查询，我们可以根据我们的输入构造查询..在Hql查询是静态查询的情况下，一旦我们构造，我们就不能改变查询的结构。