There is currently no comprehensive molecular hypothesis to account for position-effect variegation, the mosaic expression of a gene lying near a breakpoint of a chromosomal rearrangement. Here it is proposed that position-effect variegation arises from either anti-sense transcription or from promoter occlusion (transcription readthrough), the former mechanism operating for breakpoints on the 3′ side of the affected gene and the latter for breakpoints on the 5′ side. Anti-sense transcription will occur in rearrangements that place the anti-sense strand of genes next to a promoter. This anti-sense RNA hybridizes to, and thereby inactivates, sense mRNA transcripts (as anti-sense RNA is known to do). Promoter occlusion may occur in rearrangements that place the affected gene near an open upstream promoter. This promoter drives readthrough transcription that inhibits most normal transcripts. Occasional normal transcripts lead to phenotypic variegation. These hypotheses have three strengths: (i) they predict the major observed features of position-effect variegation including variegated phenotype, stable inheritance, the involvement of rearrangements, only some rearrangements causing variegation, the occurrence of both dominant and recessive variegation, the spreading effect of variegation to several loci, and the conditions required for expression of variegation; (ii) they can plausibly account for features of position-effect variegation that they do not specifically predict; (iii) they lead to a series of novel and testable predictions, including the presence of altered transcripts in rearrangements inducing position-effect variegation, the location of breakpoints required to cause variegation, and a correlation between the extent of the spreading effect and the length of the novel transcript. These mechanisms can account for several other cases of variegation in addition to classic position-effect variegation. Actual or putative examples of phenotypic variegation due to these mechanisms are known.