The Legacy American Factfinder interface at the Census Bureau website includes 1990 and 2000 Census demographic data.
Choose one of the links under Data Sets under the heading Getting Detailed Data.
NOTE: For help extracting data from the new American Factfinder, see this guide.
Summary Files (SF) 3 and 4 data have the most number of variables, including income data, although they're derived from samples, not actual counts, and don't go down to the smallest geographical level (block). SF 1 and 2 go down to the block level and the data are derived from actual counts, but they only have basic information such as race and age. SF 2 and 4 have much more detailed racial breakdowns than do SF 1 and 3.
![]() |
After choosing the "File" you want, you can browse available variables in each Data Set by clicking on About this data set. You get a list of variables with the number of possible answers in brackets (i.e., number of fields of data for that one variable). | ![]() |
The Technical Documentation (PDF) link will give you an expanded list and lots more information.
The option Detailed Tables is good for building a customized spreadsheet. You then just make choices on geographic level, then on the variables you need, then on output.
Just follow the directions one step at a time, from top to bottom. It sometimes takes a moment for the Java applet to redraw the window after each selection. When you get the geographical units you want in the final box, click Next.
To change to a different data set, click on a higher hierachical level in the "You are here" section at the top.
The geo within geo selection method is useful if you want to compare data for all of certain type of geographical unit (say, counties) that are contained within a geographical unit that is more than one hierarchical level larger (say, nation; this skips the state level).To understand the geographical levels used, go to Explain Census Geography.
Just follow the directions one step at a time, from top to bottom. NOTE: Make sure that you select one of the total population variables for the geographic level you need (e.g., for each tract) so you can later normalize other variables (e.g., get the variable as a percentage of tract population). When you get the demographic or housing variables you want in the final box, click Show Table.
To review what you will get when you choose a variable, it's recommended you click the What's this? button to the right of the list of variables. 
For instance, if you choose to download 2000 SF 3 variable P6, "Race," you will find it contains 8 fields (i.e., columns in a spreadsheet), one for each race listed. (The value for each field will be the number of persons in that geographic unit who identify themselves as a member of that race.)
The number in square brackets after the ALL-CAP variable name is the number of fields for that variable.
The Download link is at the top of your table. You have several format options (may differ for "detailed tables," "custom tables," etc.) for downloading and you will need to clean up the files before opening them in ArcGIS. The main problems revolve around getting a columns with values that can be used to match an attribute in a target attribute table in ArcGIS, in getting that match-up column defined in text format, and in tweaking the file so it retains desired features when saved to dBase.
Here's one method that works.
There is a discussion on saving from Excel to dBase format from the University of Texas School of Architecture. This only works from older versions of Excel.
In the GeoID field, use Edit...Replace... to strip the first 7 characters (14000US) and replace with a single quote ( ' ) using Edit...Replace . The resulting codes should match the codes in ESRI's STFID column. (The single quote defines the characters as text and prevents Excel from stripping the zero ( 0 ) from the beginning of the codes for states early in the alphabet, such as Arizona.)
You end up with text values in the cells of Column A, denoted by a little green triangle (Excel is "warning" you that these are numerals defined as text, which is exactly what you want!). There are other ways to define the cells or the column as "text," but the values won't save as text in dBase unless you see the little green triangles, and ArcGIS won't be able to join the table to a spatial layer.Unless otherwise specified on this page, this work is licensed under a
Creative Commons Attribution-Noncommercial-Share Alike 3.0 United States License.