Various methodological pitfalls preclude simple validation of the ostensible safety of offspring conceived thorough assisted reproductive technologies (ART). These pitfalls include lack of an appropriate comparison group (control), failure to take into account potential confounding variables, anomaly surveillance that is either too rigorous or too lax compared to the way in which anomalies are sought in general population surveys, and inconsistent criteria with respect to classifying anomalies (inclusion or exclusion of minor anomalies, internal anomalies evident only on ultrasound, and anomalies present in terminated pregnancies). To minimise these pitfalls, we recommend prospective surveillance for major anomalies, best defined as defects resulting in death, causing major handicap or necessitating surgery. Surveillance should ideally begin as soon as pregnancy is diagnosed. Data should be gathered allowing potential confounding variables to be taken into account, and systematic neonatal surveillance for anomalies should be performed during a designated time interval.