Transverse, single‐spin asymmetries offer a chance to test QCD at the level of ‘‘twist‐3’’ observables. Early suggestions that such asymmetries necessarily vanish asmq→0 or involve an extra power of &agr;scan be refuted with a simple example. Recent experimental results support the interpretation of these data in hard‐scattering QCD. The asymmetry in the scattering on nuclear targets can provide new, nontrivial information the space‐time structure of the interaction.