This paper investigates the structure of incremental liability rules. Necessary and sufficient conditions are derived for an incremental liability rule to be efficient. A liability rule, in the ordinary sense of the term, to be called a standard liability rule, is a rule which specifies the proportions in which the loss, in case of accident, is to be apportioned between the victim and the injurer as a function of their proportions of non-negligence. An incremental liability rule is a rule which specifies (i) which of the two parties, the victim or the injurer, is to be the non-residual liability holder; and (ii) the proportion of the incremental loss, which can be ascribed to the negligence of the non-residual party, to be borne by the non-residual liability holder. The necessary and sufficient conditions for an incremental liability rule to be efficient, derived in the paper, can be stated as follows: Let the party which is the residual liability holder when both parties are non-negligent be designated as r and the other party as nr. An incremental liability rule is efficient for every admissible application iff its structure is such that: (i) If party r is negligent and party nr is non-negligent, then party r must remain the residual liability holder. (ii) If party nr is negligent and party r is non-negligent, then party nr must either become the residual liability holder or liability of nr must be equal to the entire incremental loss which can be ascribed to the negligence of nr.
