Although the measurement of intracranial pressure (ICP) is gaining widespread acceptance, the most desirable method of measurement is disputed. Subdural fluid-coupled techniques are associated with an increased risk of infection, and epidural techniques are associated with decreased accuracy. We investigated epidural measurement techniques and suggest that the necessary and sufficient criteria for accurate epidural measurement of ICP are adequate transducer size and stiffness, transducer-dura coplanarity, transducer-guard ring coplanarity, complete dural contact, and rigid fixation. An epidural transducer design was developed and prototypes were constructed using these principles. The transducer requires no percutaneous connections, fluid coupling, or batteries. Transducer accuracy was ±2.2 torr in bench stability studies lasting up to 198 days, ±3.0 torr in acute animal studies of less than 24 hours, and ±7.9 torr in chronic animal studies lasting up to 112 days. Error bounds are expressed such that 95% of individual measurements are expected to have error less than the bound; average error is one-third of the bound. Average transducer drift was 0.1 torr per day; our reported accuracy in chronic studies used drift correction from preimplantation data. We conclude that accurate measurement of ICP using an epidural transducer is feasible.