Larval lampreys (ammocoetes) are assumed to be sedentary and have restricted movements, but their movement is still poorly understood. Using elastomer implants, ammocoetes of two different species of lampreys (Lethenteron appendix and Petromyzon marinus) in two streams were tagged with permanent, unique marks (N = 1035) and resampled once a month from June 2014 until October 2014. Some individuals were recaptured during every sampling event (N = 3, 0.3%). However, most animals were never recaptured after being tagged (N = 700, 67%) or were captured only once (N = 213, 21%). The apparent survival (ϕ) varied in response to time and grouping variables at both sites (0.25–0.86), but the probability of capture (P) did not vary as much (0.41–0.85). Differences were detected between sites, with ϕ being higher for L. appendix at Dyke Creek, a Genesee River tributary. Whether differences were caused by site‐specific details or species differences is unclear. At Dyke Creek, where age could be partially established from length‐frequency histograms, ϕ was higher for 1 year old than for older animals. At Oquaga Creek (where P. marinus were captured), ϕ was highly variable (0.11–0.72) and was lowest in the fall sampling (September–October). The P was similar at both sites (0.5–0.6) and across times and groups, even when it was significant to the model. Understanding the movement of ammocoetes will help conserve and control lampreys.