The objective of this study was to examine recruitment in three observational follow-up studies of patients with stroke, focusing on reasons for nonparticipation and the role of potential factors in explaining recruitment rates. It comprised secondary analysis of the three studies. Recruitment rates varied between the studies. Between 10 and 50% of those screened participated. In one study with no eligibility criteria other than informed consent being obtained and a confirmed diagnosis of stroke, we recruited only 50% (95% confidence interval 47 to 54%) of those screened. Recruitment is particularly difficult when restricted to a time window around discharge from hospital. We recommend that potential participants should be approached for consent as soon as possible, so that data tracking their hospital stay can be retained. Even where effectively no eligibility criteria are used, recruitment is limited by the necessity to gain informed consent, particularly amongst patients with severe disease.