Unir tabelas com linhas

Se você tiver apenas alguns números de posições, poderá fazê-lo com PIVOT palavra-chave

select
    UserID,
    UserName,
    [1] as Position1,
    [2] as Position2,
    [3] as Position3
from
(
    select
        U.UserID, U.UserName, P.PositionName,
        row_number() over (partition by U.UserID order by P.PositionName) as RowNum
    from Positions_Users as PU
        inner join Positions as P on P.PositionID = PU.PositionID
        inner join Users as U on U.UserID = PU.UserID
) as P
    pivot
    (
        min(P.PositionName)
        for P.RowNum in ([1], [2], [3])
    ) as PIV

SQL FIDDLE

Se, no entanto, você deseja ter um número dinâmico de colunas, deve usar SQL dinâmico, como este

declare @stmt nvarchar(max), @stmt_columns1 nvarchar(max), @stmt_columns2 nvarchar(max)
declare @Temp_Data table (RowNum nvarchar(max))

insert into @Temp_Data
select distinct row_number() over (partition by U.UserID order by P.PositionName) as RowNum
from Positions_Users as PU
    inner join Positions as P on P.PositionID = PU.PositionID
    inner join Users as U on U.UserID = PU.UserID

select @stmt_columns1 = stuff((select ', [' + RowNum + ']' from @Temp_Data for xml path(''), type).value('.', 'nvarchar(max)'), 1, 2, '')
select @stmt_columns2 = stuff((select ', [' + RowNum + '] as Position' + RowNum from @Temp_Data for xml path(''), type).value('.', 'nvarchar(max)'), 1, 2, '')

select @stmt = '
select
    UserID,
    UserName,' + @stmt_columns2 + '
from
(
    select
        U.UserID, U.UserName, P.PositionName,
        row_number() over (partition by U.UserID order by P.PositionName) as RowNum
    from Positions_Users as PU
        inner join Positions as P on P.PositionID = PU.PositionID
        inner join Users as U on U.UserID = PU.UserID
) as P
    pivot
    (
        min(P.PositionName)
        for P.RowNum in (' + @stmt_columns1 + ')
    ) as PIV'

exec sp_executesql @stmt = @stmt

SQL FIDDLE