I realize this is similar to Stuart Welsh's answer; however, I'd like to make his good answer a bit more comprehensive.
Basically, Apple recommends that you create a "world" node to contain all of your other nodes, with the world being a child of the actual scene; and further that you create a "camera" node as a child of the world node.
Then, from your scene, you can do something like:
[self centerOnCameraNamed:myCameraName];
Which calls the method (also in your scene):
- (void)centerOnCameraNamed:(NSString*)cameraName
{
SKNode* world = [self childNodeWithName:worldName];
SKNode* camera = [world childNodeWithName:cameraName];
CGPoint cameraPositionInScene = [camera.scene convertPoint:camera.position fromNode:world];
world.position = CGPointMake(world.position.x - cameraPositionInScene.x, world.position.y - cameraPositionInScene.y);
}