Variable precision rough sets model is an extension of classical rough sets theory and can deal with inconsistent data. Upper and lower distribution reduction are important attribute reduction in rough set. Utilizing distribution function, the paper puts forward some theories about upper and lower distribution reduction and appropriate binary discernable matrix based on Ziarko variable precision rough sets model. That applying logic operation in binary discernable matrix can carry out attribute reduction to variable precision rough sets model, which can solve the problem about upper and lower distribution reduction of inconsistent decision tables effectively.