The invention aims to provides a data increment processing method and system based on a distributed off-line database and aims to solve the problem of data repeating of a big data platform due to main key value alternation after data increment merging in an existing technical scheme. The method comprises the following steps that key information of increment data is obtained and is written into an increment change table; increment data records alternated in operating state in an increment data record table serve as data records to be processed; unequal records in the data records to be processed before main key update and after main key update are obtained and serve as data records to be divided; the data records to be divided is divided into data records deleted in operating state and data records added in operating state; data increment merging is performed according to an increment change table. The data increment processing method has the advantages of preventing data from being repeated due to main key alternation.