Twenty-seven narcoleptic patients severely affected with cataplexy completed four symptom diaries over a 4-month period in order to clarify some of the controversies surrounding assessment of anticataplectic medications. The home diary method was found to be a viable model for the assessment of anticataplectic activity. Assessment of reliability in 1-, 2-, 3-, 4-, 5- and 10-day intervals indicated that reliability increases with the number of days included. A 10-day design was found to be optimal. Reliability decreased, however, with each successive diary over the 4-month period. Power analysis indicates that two groups of 30-40 subjects in a parallel design, or one group of 30-40 subjects in a crossover design, would be sufficient to demonstrate a significant therapeutic anticataplectic effect in most cases. A "first diary effect" was observed, suggesting that a training period prior to the actual trial might improve reliability. Whether the patient was treated or untreated with stimulant medications did not affect severity or fluctuation of cataplexy, suggesting that both groups of patients could be included in therapeutic trials. No time-of- day fluctuation was observed in the daily distribution of cataplexy attacks. Sudden increases in cataplexy were often, although not always, caused by unusual emotional events or sleepiness. The finding of a long- lasting "precataplectic" feeling or "aura" pointed to the need to carefully clarify the symptom prior to beginning a therapeutic trial.