Current models of segregation distortion based on previous experimental results predict that, in the Sd heterozygous Rspi/Rsps male, the chromosome carrying the sensitive Rsps allele is distorted or transmitted in a frequency smaller than that of the expected Mendelian 0.5 relative to the chromosome carrying the insensitive Rspi allele. The present study presents a case where this does not occur, that is, when the genotype of the males is supp-X(SD)/Y; Sd E(SD)Rspi M(SD)+/Sd+ E(SD)+ Rsps M(SD)+ where supp-X(SD) is an X chromosome carrying a strong suppressor or suppressors of SD activity and SD+ E(SD)+ Rsps M(SD)+ is the standard cn bw chromosome. Following the "inseminated female transfer" procedure, young males of the above genotype carrying the standard-X instead of the supp-X(SD) chromosome show k values for the SD chromosome (frequencies of the SD chromosome recovered among progeny) of about 0.75, but with the supp-X(SD) chromosome, the k values are reduced to 0.36-0.41. Several possibilities other than the mechanism of segregation distortion to explain the reduced k values are ruled out. The occurrence of "negative segregation distortion" is clearly demonstrated, where the chromosome carrying the Rspi allele is distorted but the chromosome with the Rsps allele is not. This result requires a major modification of the current models or even a new model for the mechanism of segregation distortion to accommodate Rsp allele sensitivity or insensitivity. The present study also shows that males of the genotype, Sd Rspss M(SD)+/Sd+ Rspss M(SD), are almost completely sterile, but their fertility is considerably increased when SD activity is suppressed by the presence of the supp-X(SD) chromosome. This result suggests that the amount of the Sd product is not limited with respect to the interacting sites available, that is, the amount is large enough to interact with both of the Rspss alleles.