Table:
| id | productId | orderIndex | rejected | ------------------------------------------ | 1 | 1 | 0 | 1 | | 2 | 1 | 1
http://sqlfiddle.com/#!9/0196f/2
SELECT DISTINCT t.*
FROM table1 t
INNER JOIN (
SELECT productId, min(orderIndex) minIdx
FROM table1
WHERE rejected = 0
GROUP BY productId
) t1
ON t.productId = t1.productId
AND t.orderIndex = t1.minIdx;
You can start by selecting the minimum orderIndex of products that are not rejected like this:
SELECT productId, MIN(orderIndex)
FROM myTable
WHERE rejected = 0
GROUP BY productId;
Once you have that, you can join it with your original table on the condition that productId and minOrderIndex match:
SELECT m.id, m.productId, m.orderIndex
FROM myTable m
JOIN(
SELECT productId, MIN(orderIndex) AS minOrderIndex
FROM myTable
WHERE rejected = 0
GROUP BY productId) tmp ON tmp.productId = m.productId AND tmp.minOrderIndex = m.orderIndex;
My query makes the assumption that there are no duplicate (productId, orderIndex) pairs. As long as those don't exist, this will work just fine. Here is an SQL Fiddle example.