Consider:

DECLARE @Table TABLE (ID VARCHAR(3), Variable VARCHAR(10), Value VARCHAR(5));
INSERT INTO @Table (ID, Variable, Value) VALUES
('ID1', 'Variable1', 'value'),  ('ID1', 'Variable2', 'value'), ('ID1', 'Variable3', 'value'), ('ID2', 'Variable1', 'value'),  
('ID2', 'Variable2', 'value'), ('ID2', 'Variable3', 'value'), ('ID3', 'Variable1', 'value'),  ('ID3', 'Variable2', 'value'), 
('ID3', 'Variable3', 'value');

SELECT *
  FROM @Table
  PIVOT (
         MAX(Value) FOR Variable IN (Variable1, Variable2, Variable3)
        ) p;

PIVOT requires an aggregate. Given your values are all [N]VARCHAR it doesn't really matter if we use MIN or MAX. In order to have a column returned from the PIVOT it must be included in the FOR . Adding , Variable4 would include another column with NULL values, as there aren't any in the table. Removing , Variable3 would remove that column, ignoring any values present for it.

You can do a 3-dimensional join:

SELECT t1.ID, t1.value AS Variable1, t2.value AS Variable2, t3.value AS Variable3
FROM yourtable t1
JOIN yourtable t2
ON t1.ID = t2.ID
JOIN yourtable t3
ON t2.ID = t3.ID
WHERE t1.Variable = 'Variable1' AND
      t2.Variable = 'Variable2' AND
      t3.Variable = 'Variable3'

This way, you join together 3 records for each row, force t1 to be of Variable1, t2 to be of Variable2, t3 to be of Variable3. And you have no aggregation at play. You simply read all the values as they are and name them as Variable1, Variable2 and Variable3 for clarity's sake.

However, the query above assumes that for each ID you have a Variable1, a Variable2 and a Variable3. If you cannot make that assumption, then you will need to use outer joins. But, since your question's example does not suggest this being possible, this answer for the time being will assume that each tuple will have all of the 3 components existent.