Objective: To assess the relation between coffee consumption and seizure frequency in patients with drug-resistant focal epilepsy. Methods: Cross-sectional analysis of data collected in the SAVE study, which included patients with drug-resistant focal epilepsy during long-term EEG monitoring. Patients in whom both coffee consumption and data about seizure frequency, including focal to bilateral tonic-clonic seizures (FBTCS), were available were selected. Coffee consumption was collected using a standardized self-report questionnaire and classified into four groups: none, rare (from less than 1 cup/week to up 3 cups/week), moderate (from 4 cups/week to 3 cups/day), and high (more than 4 cups/day). Results: Six hundred and nineteen patients were included. There was no relation between coffee consumption and total seizure frequency (p = 0.902). In contrast, the number of FBTCS reported over the past year was significantly associated with usual coffee consumption (p = 0.029). Specifically, number of FBCTS in patients who reported moderate coffee consumption was lower than in others. In comparison with patients with moderate coffee consumption, the odds ratio (95%CI) for reporting at least 1 FBTCS per year was 1.6 (1.03–2.49) in patients who never take coffee, 1.62 (1.02–2.57) in those with rare consumption and 2.05 (1.24–3.4) in those with high consumption. Multiple ordinal logistic regression showed a trend toward an association between coffee consumption and number of FBTCS (p = 0.08). Conclusions and relevance: Our data suggest that effect of coffee consumption on seizures might depend on dose with potential benefits on FBTCS frequency at moderate doses. These results will have to be confirmed by prospective studies.