For two resistors to be in parallel then they share common nodes on both sides.
Do this:
1) Label each node, (ie A, B, C, ...)
2) If two resistors share the same "input" nodes and "output" node, then squeeze them together (combine them as an equivalent resistor) and label them as a new resistor (the parallel combination).
3) Redraw the circuit, and rinse and repeat (start at step 1 again)
We'll use these rules for your circuit.
1) We label all the nodes.
Recall: A node can be defined as a "point" where two elements branch off from.
So labeling the nodes, On the left side we have (from top to bottom).
A on the very top (with the voltage source to the right).
A in the middle (with R1 to the right).
A on the bottom (with R3 to the right).
B in the middle on the top (voltage source to the left, R2 to the right)
B in the middle in the middle (R1 to the left, R4 to the right)
C on the top right (R2 to the left)
C in the middle on the right (R4 to the left)
C on the bottom on the right (R3 to the left)
2) If two resistors share nodes, then we "squeeze" them together.
You said you can see why R2 and R4 are in parallel, so let's use the rules I defined to show that.
Notice:
R2 has the nodes B, C
ie, B----R2------C
R4 has the nodes B, C also
ie, B----R4------C
So we can squeeze them together and call them a new node, how about R2 || R4
3) Now redraw the circuit (note: you keep the input and output nodes labels the same).
Let me try to draw you a ghetto ascii circuit
A------Voltage_Source-----B----(R2||R4)------C
|.......|......|
|.......|......|
A------------R1-----------B......|
|............|
|............|
A-------------R3------------------------------C
Now which resistors share nodes?