The recognition of twinning and the determination of the twin laws are nowadays often entrusted to "black-box" software packages. This approach may be both risky and uneconomical. It is risky because never a procedure should be blindly applied. It is uneconomical if procrastinates the analysis of twinning at the very late stage of structure solution and refinement, whereas in many cases this analysis can and should be applied much earlier. In this paper we present a brief survey of the working strategies to be applied to the structural study of twinned crystals, emphasizing the role of the crystallographer, rather than that of the machine.