Don't. Unless you use diodes on the outputs, one supply will feed back into the other. If you use steering diodes you still will never be able to set the voltages close enough to equally divide the load between the two supplies. The one with the higher voltage will supply most of the load current.
Some supplies are designed to be paralleled and have a tracking connection between them. Unless you have one of these it is best not to parallel two regulated power supplies.
I know this sounds argumentative, but that's not my intention. I see this answer given quite regularly and I know that it's wrong. For instance, we parallel batteries all the time and nothing bad happens. We do this with regulated power supplies all the time, too. Ones that don't have the sensing connection between them are used more often than ones that do. I'd really like to see an explanation that includes a proof of why paralleling power supplies is different.
Here's my explanation... Assume that we just have two constant voltage power supplies, where V0 = V1. We connect the ground terminals together and we connect the positive terminals together. We put a load that draws current I. I like the nodal method to analyze problems like these, so that's what we'll do.
Pick a reference node from which we will measure voltages. Ground is nice for that. We only have one node left and we'll call it e. Now we write KCL (Kirchoff's current law) for the node.
(e-V0) + (e-V1) - I = 0. We can see by inspection that e=V0=V1 and that I = i0 + i1. So i0 has to be the same as i1, when the voltages are equal. If V0 ≠ V1, then i0 ≠ i1, but only by the amount the voltages differ.
If I'm doing this wrong, please let me know where I went astray. I'd like to correct my errors.