Physicians are traditionally liable under a negligence rule of liability. Economic analysis of liability rules, including malpractice, assumes that the primary function of liability is injury prevention (deterrence). Compensation can be provided more efficiently through other forms of social or private insurance. In theory, a negligence rule creates incentives for efficient care, hence there should be no negligence, no claims and no demand for liability insurance. In practice, the incidence of negligent injury has been estimated at roughly one per hundred hospital admissions in the US and about one in seven physicians is sued per year.These discrepancies between the theory and actual operation of the negligence system arise primarily because of imperfect information on the part of courts, doctors, patients, liability insurers and health insurers. Imperfect information and extensive health insurance lead to biased and uncertain legal standards. Uncertain legal standards create incentives for physicians to practice defensive medicine and incentives for plaintiffs and defendants to invest in litigation, leading to high overhead costs, such that compensation through the malpractice system carries a load of $1.50 per $1.00 of compensation.Nevertheless, the extreme criticisms of the malpractice system are exaggerated. Malpractice premiums are less than 1 percent of total health care costs. There are no comprehensive estimates of defensive medicine costs; in any case such costs are likely to decline with the growth of managed care. Although claim disposition exhibits both Type 1 and Type 2 errors, negligent injuries are much more likely to lead to a claim being field and payment to the plaintiff than non-negligent injuries, and awards are strongly related to loss incurred. The limited empirical evidence of provider response to liability and the deterrent effect of claims suggests -- but cannot prove -- that the net benefits of the malpractice system may plausibly be positive. Nevertheless, reforms designed to reduce inappropriate compensation and deter excessive litigation and defensive practice would make the system more cost-effective.The empirical evidence, based primarily in the US, includes studies of malpractice injuries; physician response to liability; trends in claim frequency, severity (size), and claim disposition; and the malpractice insurance market. Analyses of actual and proposed reforms address tort reform, no fault, enterprise liability and optimal liability under managed care. More limited evidence is available on the negligence regimes in Canada and the UK, and the quasi no-fault regimes in Sweden and New Zealand.