The first three exons of the human muscle dystrophin gene were expressed as a beta-galactosidase fusion protein. This protein was then used to prepare two monoclonal antibodies (mAbs) which react with native dystrophin on frozen muscle sections and with denatured dystrophin on western blots but which do not cross-react with the dystrophin-related protein, utrophin. Both mAbs recognized dystrophin in muscular dystrophy (MD) patients with deletions of exon 3, and further mapping with 11 overlapping synthetic peptides showed that they both recognize an epitope encoded by the muscle-specific exon 1. Neither mAb recognizes the brain dystrophin isoform, confirming the prediction from mRNA data that this has a different N-terminus. One Becker MD patient with a frameshift deletion of exons 3-7 is shown to produce dystrophin which reacts with the N-terminal mAbs, as well as with mAbs which bind on the C-terminal side of the deletion. The data suggest that transcription begins at the normal muscle dystrophin promoter and that the normal reading frame is restored after the deletion. A number of mechanisms have been proposed for restoration of the reading frame after deletion of exons 3-7, but those which predict dystrophin with an abnormal N-terminus do not appear to be major mechanisms in this patient.