Thyroid cancers detected by screening irradiated individuals are often small and of uncertain clinical significance. We retrospectively analyzed the effect of screening in a cohort of 4296 individuals exposed to radiation as children in the 1940s and 1950s and followed by us from 1974 until the present. We compared the thyroid cancers diagnosed before 1974 (122 cases, routine care) with the cancers found in subjects screened by us after 1974 (172 cases, screened), using cancer recurrence as the end point. Screening included a thyroid scan or, more recently, thyroid ultrasound. As expected, many of the cancers found by screening were very small (52% were <10 mm), but the range of tumor sizes overlapped those found by routine care. The recurrence rate was significantly lower in the cases found by screening, but when the comparison was limited to cancers 10 mm or larger, no difference in the recurrence rates was seen. This would suggest that the lower recurrence rate observed for small thyroid cancers detected at screening was due to earlier diagnosis rather than more effective treatment. By univariate analysis, four factors were associated with an increased risk of recurrence of small (<10 mm) thyroid cancers: short latency (i.e. a shorter time interval between the radiation exposure and the first thyroid surgery), lymph node metastases present at diagnosis, multifocal cancers, and higher radiation dose. In a multivariate analysis combining the four risk factors, only short latency was significant. As thyroid cancers that escape detection by routine means should be diagnosed at screening, and both large and small thyroid cancers have the potential to recur, screening may be of value, but only if groups with a sufficiently high prevalence of thyroid cancer can be identified to offset the adverse effects of unnecessary treatment due to false positive results.