你忘了将子查询限制在当前的人，但是，如果这是我的选择，我会用CTE写它，并加入它：

with visit2017 as 
   (select sum(events) as sum_events, PersonID
    from Visit 
    where date between 20170101 and 20171231 
    group by PersonID ), 
visit2016 as
   (select sum(events) as sum_events, PersonID 
    from Visit 
    where date between 20160101 and 20161231 
    group by PersonID ) 
select distinct 
       person.personid, 
       person.DOB,
       person.FullName, 
       visit2017.sum_events as visit2017,
       visit2016.sum_events as visit2016
into #temp
from table1 person 
left join table2 visit --< I don't understand this joing
on person.personid=visit.personid
left join visit2017 
on person.personid=visit2017.personid    --< important
left join visit2016
on person.personid=visit2016.personid    --< important
where visit.date between 20160101 AND 20171231 ;

在子查询和外部查询中使用别名并添加where子句。

SELECT PersonID,
       ( SELECT sum(events) from Visit AS V where date between 20170101 and   
         T.PersonID = V.PersonID ) AS VV
FROM TableA AS T

从您的查询看，它如下所示：

select sum(events) from Visit where date between 20170101 and 20171231 group by PersonID

可以返回多个记录，因为它为每个personID返回总和（事件）。请尝试单独运行此程序，以查看它是否返回多个结果

出现该错误的原因是子查询希望返回1个结果，但查询生成了多行。您可以自行运行子查询以确认错误。
您可以将子查询重构为连接

select distinct 
person.personid, 
peron.DOB,
person.FullName, 
sum(visit.events) AS visit2017, 
sum(visit2.events) AS visit2016

into #temp
from table1 person 
left join table2 visit
    on person.personid=visit.personid
    AND date between 20170101 and 20171231
left join table2 visit2
    on person.personid=visit2.personid
    AND date between 20160101 and 20161231
where (visit.date between 20170101 and 20171231
OR visit2.date between 20160101 and 20161231) 
GROUP BY    
person.personid, 
peron.DOB,
person.FullName;

当你这样做的时候，通常最好是在一次选择中计算所有的年份，这样就不必多次读取表，而且如果你使用cross apply，你也不需要将结果乘以所有的事件，也不需要使用distinct。

select 
    person.personid, 
    person.DOB,
    person.FullName,
    visit.visit_2017,
    visit.visit_2018
into 
    #temp
from 
    table1 person
    cross apply (
        select 
            sum(case when date >= '20170101' and date < '20180101' then events end) as visit_2017,
            sum(case when date >= '20160101' and date < '20170101' then events end) as visit_2016
         from 
             Visit 
         where
             date >= '20160101' and date < '20180101' and 
             Visit.PersonID = person.PersonID
    ) visit

它看起来要复杂得多，但并不难学，Case inside a sum通常很有用，使用outer apply你可以从同一个查询中得到多个结果，并将其放入最终的select中，这与select部分中的（select..）不同。
我还将between改成了两个独立的〉=和〈。Between在使用日期时可以正常工作，但如果是datetime/smalldatetime等，那么在时间不是00:00的最后一天，您将错过任何内容

我假设这里有一个表Visit，其中Date、PersonID是字段，还有一个表person，其中personid、DOB和FullName是字段。

select person.personid,
       person.dob,
       person.fullname,
       Count(Year(visit.date))
from   person p
       join visit v
         on p.personid = v.personid
where  Year(visit.date) in ( 2016, 2017 )
group  by p.personid,
          p.dob,
          p.fullname,
          Year(visit.date)

这将计算每年的访问次数