Systematic errors specific to a snapshot Mueller matrix polarimeter are studied. Their origins and effects are highlighted, and solutions for correction and stabilization are proposed. The different effects induced by them are evidenced by experimental results acquired with a given setup and theoretical simulations carried out for more general cases. We distinguish the errors linked to some imperfection of elements in the experimental setup from those linked to the sample under study.