Assessment of risk to benefit ratio in patients with epilepsy is crucial in determining the need for treatment, the choice of drugs and the use of monitoring tools such as laboratory tests and other investigations. Active epilepsy per se carries significant risks in terms of increased mortality, susceptibility to psychopathology and physical injury, and reduced quality of life as a result of restricted lifestyle, stigma and prejudice. By preventing the occurrence of seizures, antiepileptic drugs (AEDs) attenuate or eliminate altogether seizure-related risks, but other risks may arise due to the side effects of the drugs, all of which have a relatively narrow therapeutic index. While there are no major differences in the degree of efficacy between AEDs which are effective in any given seizure type, side effect profiles differ considerably from one agent to another and represent a major factor in determining choice of treatment. Assessment of risk to benefit ratio should also take into consideration patient-specific factors such as type and severity of the epilepsy, age, sex, childbearing potential, medical and drug history, associated disease, use of concomitant medication (including the contraceptive pill) and the prospected patient's compliance. In some benign epilepsy syndromes, such as idiopathic partial epilepsy with centro-temporal spikes, the risk of side effects from AEDs may outweigh potential benefits in terms of seizure control, and treatment is generally not indicated. At the opposite end of the spectrum, the serious morbidity and mortality associated with severe epileptic encephalopathies, such as the Lennox-Gastaut syndrome, justifies aggressive treatment even with drugs associated with a relatively high risk of life threatening side effects such as felbamate. The present article will provide an overview of specific risks associated with epilepsy and with the various drugs used for its treatment, and will attempt to evaluate the complex balance between these risks and therapeutic benefits in different categories of patients.