1. The causal relationship of time of day, sleep and other factors to the frequency of major convulsive states has been studied in 31 epileptics over a period of one year.2. It is found that they fall into three groups—namely, those in whom the attacks are predominantly diurnal, mainly nocturnal, or occur indifferently by day or night.3. Attacks are more frequent in the nocturnal and diurnal groups than in the diffused group.4. Approximately two-thirds of 31 epileptics manifested a marked difference in the diurnal and the nocturnal incidence of their convulsions, and approximately two-thirds exhibited one or more "time peaks," i. e., a concentration of attacks at or around certain hours.5. Nine such time peaks can be recognized, four diurnal, three nocturnal and two diffused. Their interpretation is discussed.6. Patients of the "diurnal" and "nocturnal" groups are found to respond in an opposite manner to the onset and cessation of sleep.7. There appears to be a basic distinction between the patients composing the respective groups. The distinction seems to depend on many factors—endogenous and exogenous—important among them being the type of sleep and associated conditions, transition from waking to sleep and vice versa and consequent postural and circulatory changes, alveolar CO2 tension and temperature variations.8. Many patients show a fairly definite regular periodicity of fits, not only with respect to day, but also the day of the month, and the month of the year and season. Absence of the expected occurrence of fits, or irregularities should lead us to seek causes for the change and might result in a decrease of the number of fits through ascertaining modifiable or preventable factors at work. These factors are often situational or dietetic.9. The pooled fits of all the patients of the three groups show a greater seasonal incidence during the spring months. This might be partially due to hyperventilation and hyperparasympatheticotonia.10. Four patients suffered from organic lesions of the nervous system. Most of these belonged to the diffused group, but in other respects the type of their seizures differed in no essential from the 27 cases of "idiopathic" epilepsy.11. A larger percentage of the diffused group craved salt, milk or meat than of the other two groups. On the other hand various forms of excitement precipitated more fits in the diurnal and nocturnal types. In the latter group there was a higher percentage of mental instability than in the diffused and diurnal groups.12. There is little difference in the mental age of the three groups, although mental deterioration seems to be less rapid in the diurnal type, a fact which might be related to a lesser degree of mental instability in this group.13. Asthenic and dysplastic physical constitutional types predominate in the diurnal and diffused groups, whereas the pyknic type was more frequent in the nocturnal group.14. A high systolic blood pressure and particularly a high pulse pressure were found to be poor prognostic signs.15. Individuals in the diffused group are more likely to have a longer psychotic period than those in the other two groups, notwithstanding the older average age of the onset of the convulsive state in the individuals of the former group. The latter group is more likely to include the non-idiopathic types.16. Patients showing isolated single fits are more numerous in the nocturnal and diffused group. The nocturnal type seem more prone to status epilepticus.17. The diurnal group shows the longest interval between attacks, and the diffused group, the shortest. This might account for the relatively minimal deterioration level in the diurnal group.18. There is a tendency for the number of fits as well as the time duration of the auræ to diminish with the increased duration of the disease.19. The average duration of the post-convulsive stupor and confusion is somewhat longer in the diurnal group.20. The regular occurrence of nocturnal fits and to a somewhat less degree diurnal fits is usually a poor prognostic sign.21. The generally held view that the younger the individual at the time of onset of the convulsive state, the longer the duration of the period of fits, would seem to be confirmed by our findings.22. A few remarks concerning the nature of the epileptic disorder and its treatment are discussed.