I am focusing on 1850-1940 with this review.
 1900 introduces the month and year of birth - double check it! The info could be wrong.
 1900 and a couple of others list children born/living. Check your list of kids to see if you overlooked anyone.
 Immigration status: Year of immigration may be wrong. [my gg-grandmother shows 3 different years. Foriegn women who married US citizens were granted citizenship, so avoided naturalization process.]
 Birthplace for individual & parents: It could vary from census to census. Foreign birthplaces could be general or specific [i.e.: Germany one year, Baden the next] You could also draw a blank or 'unknown.' Check all censuses available for the birthplace of your ancestor.
 Age at last birthday: Nobody lies about his/her age, right? Confirm! [Example: 23 in 1850, 28 in 1860, 32 in 1870.] If you don't know the month of birth, you might want to go with two years. [45 in 1850: could have been born in 1804 or 1805; put 1804/5 or c1804, c1805, etc.]
 Confirm that sons/saughters arenatural and not step. The records generally showed 'stepchild', but not always.
 Some stepkids were listed with stepfather's surname rather than thier birth surnames, make sure of relationships.
 Accuracy depended on who gave the info to the enumerator. If no one was at home, a neighbor might have been queried. Names, ages & other details given could be questionable.
 From year to year, names could vary. If a daughter was named Nancy, Nan, Nancy, Ann, Annie could have been given. Mary Louise could have been Mary in 1860, Louise in 1880 and Lou in 1900.
 For 1850-70, don't assume relationships based on order. Since no relationships are given, mistakes can easily be made. For one of my families, researchers had assumed that a spouse of one of the sons was a sister in 1850-60.
 Not everyone in the family is necessarily going to be at home.
[a] a child could have been apprenticed to a neighbor or married sibling.
[b] the mother might have moved in with a pregnant daughter to help out.
[c] the father or a son might have gone to find work elswhere, been in the military, etc. and
returned for the next census.
 Families may have been on the move or simply missed or left out by the enumerator. Look for them the next time around. [Then again, missing from three censuses?! My James & Mary Crail are noshows in 1850-60-70. Yeah, I know, old stry! :)-]
 Spelling can be an issue. Due to poor handwriting and sloppy indexing, people can be overlooked. Get creative with spelling. Look for unusual first names. Check known neighbors and relatives, in case they are there, but didn't get indexed. If your family was in, say, Kentucky in 1850 and Ohio in 1870, but missing in 1860, check Indiana.
Confirm all you find in the census, it ain't perfect!