A clinical study was conducted to measure the effectiveness and operational characteristics of a new device intended to reduce the incidence of falls in elderly patients at high risk of falling. The NOC.watch device consists of a credit-card size device contained within an adhesive "patch"worn on the thigh continuously for many days. The patch is small, wireless, disposable, waterproof, shockproof, and unobtrusive. When a patient's leg becomes weight-bearing (such as when a patient gets out of bed or stands up unassisted), the receiver emits an audible signal which both alerts the patient to sit down and also summons a caregiver. A crossover design was used to compare the fall rate of patients while wearing the NOC.watch device with the fall rate while not wearing it. Patients (n = 47) wore the device for one week, and were monitored for fall activity before, during, and after this week. The three time periods were referred to as the "Pre," "During," and "Post" periods, respectively. The study was conducted between November 1, 2000, and April 1, 2001, in the Medicare unit of a skilled nursing facility in northern California. Forty-seven patients admitted to the unit with a fall risk assessment (FRA) score of at least 6 were entered into the study, with a mean FRA of 11.4. The main outcome measure was fall rate per 100 patient days. Qualitative observations of adverse effects on skin integrity, degree of staff and patient acceptance of the system, and lack of false alarms are also reported. Poisson regression with generalized estimating equations was used for analysis of fall rates. Forty-seven patients wore the device. Total number of patient-days of observation were 273, 299, and 497 in the Pre, During, and Post periods, respectively, for a total of 1,069 days of observation. These patients had 11 falls in the Pre period (4.0 falls per 100 days), 1 fall in the During period (0.3 falls per 100 days) and 17 falls in the Post period (3.4 falls per 100 days). This 91% reduction in fall rate while wearing the NOC.watch is statistically significant with P = 0.02. False alarms were qualitatively reported to be low or nonexistent, and patient and caregiver acceptance of the device was reported to be high. No adverse effects on skin integrity were noted. Cognitively impaired patients, who have difficulty remembering to call for assistance, appeared to respond particularly well to the device alarm. The prototype NOC.watch device appears to have a large impact on reducing fall risk in nursing home patients, with minimal adverse effects. However, due to the lack of an equivalent control group, these results should be confirmed with a larger, randomized, controlled study to better estimate the true magnitude of the effect of the NOC.watch device on fall rates.