114 patients suffering from generalized grand mal seizures were treated with phenytoin only. Serum-phenytoin was monitored, and its relation to the frequency of attacks was established in each case, observing steady state periods of at least 3 months’ duration. In a group of 44 patients the reduction in pretreatment frequency of attacks was used to evaluate the drug effect. The probability of attacks followed an exponential function of serum-phenytoin. Curves of equal probability for further attacks were plotted; they may assist in determining the initial regimen. However, the high variability in responsiveness during a serum-phenytoin below 10 mg/l suggested that the effective level has to be determined separately and repeatedly for every single patien