Closure of patent foramen ovale (PFO) is expected to prevent paradoxical emboli. In the absence of randomized trials, its efficacy has been assessed by comparing uncontrolled cohort studies of medically treated patients with those treated by PFO closure. The objective of this study was to highlight a confounder of such studies, namely, the variability in the duration of follow-up. We searched the literature for cohort studies of patients with ischaemic strokes, including those with PFO. During the first year of follow-up, recurrence hazards in patients younger than 55 years were 1-4% in those with any ischaemic stroke, 1-6% in medically treated patients with PFO and 0-5% in those after PFO closure. In most studies, the recurrence hazards were highest immediately after the index stroke and declined thereafter. Still, hazards were commonly reported in terms of annual averages over a wide range of follow-up periods for the various cohort studies, thereby ignoring the possibility that the duration of the follow-up may in and of itself affect the derived average recurrence hazards. A disregard of the time variance of stroke recurrence may confound the conclusions from comparisons between uncontrolled studies of patients with stroke and PFO.