The　problem　of　unknown　input　estimation　is　considered　in　this　paper.　The　problem　can　be　viewed　as　a　two-player　zero-sum　dynamic　difference　game.　A　game-theoretic　approach　that　incorporates　maximum　principle　arguments　is　adopted　to　solve　this　problem.　The　unknown　input　estimation　is　derived　in　terms　of　one　discrete　difference　Riccati　equation　which　is　of　a　form　suitable　for　recursive　computation　in　online　applications.　Copyright　©　2002　IFAC.