A nonsense mutation adds a premature stop signal that hinders any further translation of a protein-coding gene, usually resulting in a null allele. To investigate the possible exceptions, we used the DMD gene as an ideal model. First, because dystrophin absence causes Duchenne muscular dystrophy (DMD), while its reduction causes Becker muscular dystrophy (BMD). Second, the DMD gene is X-linked and there is no second allele that can interfere in males. Third, databases are accumulating reports on many mutations and phenotypic data. Finally, because DMD mutations may have important therapeutic implications. For our study, we analyzed large databases (LOVD, HGMD and ClinVar) and literature and revised critically all data, together with data from our internal patients. We totally collected 2593 patients. Positioning these mutations along the dystrophin transcript, we observed a nonrandom distribution of BMD-associated mutations within selected exons and concluded that the position can be predictive of the phenotype. Nonsense mutations always cause DMD when occurring at any point in fifty-one exons. In the remaining exons, we found milder BMD cases due to early 5' nonsense mutations, if reinitiation can occur, or due to late 3' nonsense when the shortened product retains functionality. In the central part of the gene, all mutations in some in-frame exons, such as in exons 25, 31, 37 and 38 cause BMD, while mutations in exons 30, 32, 34 and 36 cause DMD. This may have important implication in predicting the natural history and the efficacy of therapeutic use of drug-stimulated translational readthrough of premature termination codons, also considering the action of internal natural rescuers. More in general, our survey confirm that a nonsense mutation should be not necessarily classified as a null allele and this should be considered in genetic counselling.