All censuses are subject to some undercount. In most cases some estimate of the amount is made based on a follow up survey or comparison with other sources but in many countries it may be extremely difficult to estimate with any accuracy.
Undercount may occur for many reasons including deliberate avoidance by the population, while in some countries deliberate over-counting may be an additional obstacle to accuracy. The history of the Nigerian Census illustrates both problems. The first national census in 1952 is widely perceived to have involved an undercount of up to 10% of the population, blamed on a range of factors including fear among the population that the census was related to tax collection and logistical difficulties for enumerators in reaching the many remote areas. By contrast the accuracy of the first census in post-Independence Nigeria in 1962 was widely seen to be effected by a major problem of overcounting in many areas, essentially for political reasons (regional allocation of resources being related to population size). Since then the Nigerian census has been subjected to continual controversy Results of the 1973 were cancelled, and the only other completed count was in 1991. There have since been two unsuccessful attempts at a census, though a new census is proposed for 2005.
Problems of undercount are not limited to developing countries. Many European and Western countries, including the US have faced a growing problem of undercount over recent decades. The 1990 US Census is estimated to have missed almost 5 million people, much more than in the previous census in 1980.