Floating bird eggs to estimate their age is a widely used technique, but few studies have examined its accuracy throughout incubation. We assessed egg flotation for estimating hatch date, day of incubation, and the embryo's developmental age in eggs of the American Avocet (Recurvirostra americana), Black-necked Stilt (Himantopus mexicanus), and Forster's Tern (Sterna forsteri). Predicted hatch dates based on egg flotation during our first visit to a nest were highly correlated with actual hatch dates (r = 0.99) and accurate within 2.3 ± 1.7 (SD) days. Age estimates based on flotation were correlated with both day of incubation (r = 0.96) and the embryo's developmental age (r = 0.86) and accurate within 1.3 ± 1.6 days and 1.9 ± 1.6 days, respectively. However, the technique's accuracy varied substantially throughout incubation. Flotation overestimated the embryo's developmental age between 3 and 9 days, underestimated age between 12 and 21 days, and was most accurate between 0 and 3 days and 9 and 12 days. Age estimates based on egg flotation were generally accurate within 3 days until day 15 but later in incubation were biased progressively lower. Egg flotation was inaccurate and overestimated embryo age in abandoned nests (mean error: 7.5 ± 6.0 days). The embryo's developmental age and day of incubation were highly correlated (r = 0.94), differed by 2.1 ± 1.6 days, and resulted in similar assessments of the egg-flotation technique. Floating every egg in the clutch and refloating eggs at subsequent visits to a nest can refine age estimates.