Effective coding is critical to data collected by birth defects surveillance programs because subsequent use of the data depends on storage and retrieval of cases using codes. Hence, careful consideration needs to be given to the coding process. The primary goal of coding is to accurately, completely, and concisely represent infants with birth defects. Coding procedures need to accommodate the objectives of the surveillance program; for example, programs that focus on research may require different coding procedures from those that focus on linking infants to services. Several challenges exist in coding birth defects, including the need to distinguish infants with multiple defects and syndromes from those with isolated defects, and the need for strategies to code suspected defects for which confirmation is not available. Selection of a coding system by a birth defects surveillance program is central to the utility of the data collected. Most programs use a modification of the International Statistical Classification of Diseases and Related Health Problems-based (ICD) systems. This paper addresses ICD-based systems and the modifications used by many birth defects surveillance programs and presents examples of the problems in interpreting birth defects data because of inappropriate coding.