I describe a simple new method for collimating a Cassegrain telescope that gives near-perfect correction over a wide field of view. This method was used to collimate the 6.5 m MMT in 2002 with remarkably good results. The first step is to point the primary mirror accurately at an object. When the M1 optical axis points exactly toward an object, it cannot contribute pointing error, coma, or anamorphic aberration to the image field. If these are present, they must come from M2. Adjusting M2 to correct both pointing error and coma yields perfect collimation, canceling off-axis anamorphic aberrations. I present an analytical basis for this method and compare it with methods that measure off-axis image aberrations directly. In most cases, the technique described here gives better results. I describe the collimation procedure we used at MMT and suggest an easier way to adapt this method to other telescopes.