Conditions for and methods of aligning a two‐crystal spectrometer are discussed. These include the alignment of rotation axes, crystals, and x‐ray beam. The results of a new nondispersive crystal alignment procedure are compared in detail with theory for a typical case of CuK&agr; radiation and calcite crystals. This method is contrasted with other procedures which lead only to pseudo‐alignments.It is shown that x‐ray methods which are sensitive to vertical divergence do not lead to crystal alignment unless the beam is already perfectly aligned. It is concluded that the x‐ray beam can be aligned onlyafterthe crystals have been aligned.